Perhaps in the future, athletes will lift only with variable-resistance devices that perfectly adjust the amount of tension for the set and rep range and training goals. The first attempts at this technology are out there now, but their expense and specialization means that nearly all athletes are lifting with barbells, dumbbells, kettlebells, and other fixed-load free weights. This presents a challenge when selecting the best weight for each exercise, set, and rep. Coaches often spend a lot of time just advising athletes to increase or decrease weight, or terminate a set early based on an optimistic weight selection, taking time away from building relationships, coaching the movements, and doing all of the other things we could be doing. The two main methods of selecting training weights are percentage of one-rep max (1RM) and rate of perceived exertion (RPE). Math versus feel, which one wins?

Overall, I prefer using RPE for rowing training. It largely removes the requirement to do rep-max testing to dial in training weights, offers greater flexibility within each training session to adjust around fatigue from rowing training, and trades a lot of time spent doing math with time spent communicating and educating. However, RPE isn’t perfect, and we should be aware of the limitations, as well as the situations in which a percentage-based system may offer greater utility. In this article, I’ll break down some of the research on strength training with these two major methods, identify pros and cons of each, and provide some takeaways for how you can use RPE or percent 1RM to guide your strength training for rowing.

rpe for rowing training

Using Percentage-Based Strength Training

Percentage-based lifting is typically done as a percent of a 1-repetition maximum (1RM). This means we must do some sort of rep-max testing in order to establish a 1RM, off of which we then calculate our training weights. We can use a 2-5RM and then use a calculation to project that into a 1RM. This allows us to avoid some time, energy, and risk of going to a true 1RM, but the calculation introduces error. Many factors can influence the weight or reps that an athlete achieves in a rep-max test, from physical factors such as muscle fiber type proportion or technical efficiency, to mental factors such as level of arousal for the rep-max attempt, to emotional factors such as comfort in the training environment. For some athletes, that 5RM might be 92% of their 1RM, or 85%1RM for others. Research indicates that the greater the number of reps, the less accurate the 1RM prediction, so best practice for extrapolation is a 2-5RM.

We then take our real or projected 1RM and use it to calculate training weights and create a program progression. This introduces another area of potential error. The so-called “textbook correlations” of percent 1RM to number of sets and reps cannot be true for every single athlete. Physical, mental, and social/emotional factors can influence performance, as can interference from sport training and life stress outside of training. Using a percent 1RM system requires a belief that these correlations are close enough for our training purposes, and a willingness to accept that some athletes will be training too hard or too easy.

Where do these “textbook correlations” even come from? Loading charts like the one at right (click to open full size) are ubiquitous in weight-rooms and training resources. This one is distributed by the USA National Strength and Conditioning Association (NSCA), and was adapted from a 1984 piece that seems to only exist in the citation sections of later articles. I cannot find the full text, or even an abstract, of the original article. This loading equation was one that researchers investigated in, “Accuracy of Seven Equations for Predicting 1-RM Performance of Apparently Healthy, Sedentary Older Adults” (2009). The researchers cite personal communication with the original author, “J.” Lander, who claimed that the chart was published without his knowledge, based on his “guess-timated” calculations. Furthermore, they find that only one of the seven models they evaluated was even peer-reviewed; a 1992 study by Mayhew et al., in which the researchers studied only the bench press exercise and don’t claim validity for other lifts. The rest of the equations were simply published as charts, and then subsequently cited as though reviewed, or published based on unpublished data. This is sounding similar to the issue of “data-driven” strength standards.

And we have some idea where those come from…

rowing strength standards

Here’s why this is a problem. In “The Relationship Between The Number Of Repetitions Performed At Given Intensities Is Different In Endurance And Strength Trained Athletes” (2014), authors Richens and Cleather compared eight college-age male weight-lifters with eight college-age male runners from track and cross-country running events of at least 800 meters. They tested each athlete in a leg press 1RM, and then tested max reps at 90%, 80%, and 70% in random order on three different days separated by over 48 hours. They found that the weight-lifter group had higher 1RMs (335kg vs. 188kg), but that the endurance training group performed significantly more repetitions at 70%1RM (40 vs. 18) and 80%1RM (20 vs. 12), and non-significantly more at 90% (11 vs. 7). Given these findings, an endurance athlete doing 10 reps at 80%1RM would be severely undertraining compared to a strength-trained athlete doing 10 reps at 80%1RM.

Similarly, Desgorces et al. (2010) studied four different populations of male athletes and their performance on the bench press. 31 racket/ball players, 22 powerlifters, 28 swimmers, and 29 rowers, all with at least four years of training in their sport, did a 1RM bench press test to failure, and then two separate sessions of rep-max testing. They tested 85%, 60%, and 20%1RM in the first session, and then 75% and 40%1RM in the second session, with 15 minutes of rest between each rep-max attempt. The researchers found that all athletes performed a similar number of repetitions at 75% and 85%1RM, but the rowers and swimmers performed significantly more reps at 20, 40, and 60%1RM than the powerlifters and racket/ball players. See the graph below for a visualization of how extreme these differences became after 75%1RM.

Note: I reviewed this study in the September, 2020 issue of Science of Rowing.

rowing percent 1RM endurance versus power athletes

We don’t know to what extent this difference is due to physical, mental, and/or emotional factors. Richens and Cleather speculate that the endurance athletes may have been unfamiliar with the 1RM testing protocol, while the strength athletes had great experience in the physical, mental, and emotional skills to get the most muscular output possible. Or, perhaps both groups performed at their potential in the 1RM test, and the endurance athletes outperformed the strength athletes in the lighter weight, higher rep sets due to their increased familiarity with pushing through muscular fatigue and pain. Or, perhaps there is a physical difference in fiber proportion, oxygen transport, or other factors influencing performance for each group. Regardless of the exact mechanism, these studies together suggest that using the same percentage-based programs with different athletes would result in one group significantly over-training or under-training.

RPE-Based Strength Training

The other major approach to selecting training weights is Rate of Perceived Exertion (RPE). RPE training aims to get around the limitations of a fixed-percentage program by providing a linear scale, traditionally either 1-10, 6-20, or 1-20, in which the low numbers indicate low exertion and the high numbers indicate maximal exertion. RPE was originally introduced in the medical and scientific fields to quantify exertion in physical activity studies. It has since been adapted to correlate with heart rate for sport training, and “reps in reserve” for strength training. The goal of RPE is autoregulation, providing tools for the athlete to assess their fatigue and readiness to train, adjust their weight selection accordingly, and train in the most accurate “challenge zone” possible.

rpe for rowing trainingIn an RPE-based strength training program, the coach may program exercises, sets, reps, and rest intervals just like a percentage-based program, but instead of providing a percentage of 1RM for intensity, writes an RPE value. I’ve provided one example of a strength training RPE scale at right, adapted from Helms et al. (2018).

Using RPE, we might write, “front squat: 5 sets of 5 reps with 120s rest at RPE8.” This would instruct athletes to select their weights on each set so they feel that they have approximately two repetitions left in reserve. Theoretically, this is more accurate on a day-to-day basis than writing “5×5 with 120s rest at 87%1RM.” For 87% to be correct, we have to trust that the test conditions were accurate, that the percent 1RM correlation charts are accurate, and that one intensity is suitable for all five sets. If we want to refine our programming to adjust for the fatigue of successive sets, we might need to write “5×5 with 120s rest at 87%, 85%, 83%, 81%, 79%,” which then introduces the assumption that that fatigue loss is correct and will result in appropriate challenge for the lifter. Some programs get even more detailed with using fractions of percents. This introduces another coaching challenge, as the coach must then make sure that athletes have done the calculations and are adjusting their weights from set-to-set accordingly and accurately. Theoretically, this is already “baked in” to the RPE system. RPE8 is RPE8, whether it’s the first set or the last set, and athletes should be adjusting the weight to hit the prescribed intensity range.

RPE and Percent 1RM: Pros and Cons

On a percentage-based system, we must test rep-maxes every so often in order to calculate training weights, and we must invest planning and effort into making the test conditions accurate and reliable. Different test conditions will affect maximum muscular output. This is very challenging when working with athletes, who can have highly varying energy levels for strength training based on their sport training stress, other life stress, and familiarity with strength training exercises and test protocols. We would expect a strength athlete like a powerlifter or Olympic weightlifter, who train the same exercises in similar formats year-round with great experience in maximal testing and minimal interference from other forms of training, to perform consistently in retest conditions and succeed on a percentage-based system. I have not found as much success applying this methodology to non-strength athletes with greater variation in exercise familiarity, max testing experience, and high interference from sport training.

If we do feel that our tests and retests are accurate, then we must trust that the percentage-to-reps correlations are accurate and applicable, and hold up under rowing training stress and other life stress. If you’re a full-time athlete with nothing to do but train and recover, perhaps this is feasible. Most rowers are also students, employees, and/or responsibility-bearing family members, the stress of which can affect maximum muscular output on a daily basis. Finally, we must dedicate coaching time to spreadsheet development or doing a lot of math for the athletes we coach, and then must ensure that athletes are accurately using the calculations and adjusting their weight selections. After all of this, we’ll still have to introduce estimation strategies for programming on exercises for which we don’t have 1RM numbers. This seems like a lot of drawbacks for something that was supposed to be a simple mathematical solution to the weight-selection problem.

RPE seems to offer some major advantages over percentage-based strength training. First, we don’t need to have accurate 1RMs for every exercise. It is unfeasible and unsafe to test 1RMs on every single exercise in a training program. Therefore, there will always be some element of weight-selection estimation, particularly for assistance exercises like dumbbell press variations, horizontal row exercises, vertical pulldowns, etc. RPE offers us a theoretical system of refining this estimation process. We can use RPE for assistance work and main work alike. RPE also removes the burden to continually retest rep-maxes for the purpose of calculating training weights. RPE8 is RPE8 as long as you have two reps left in reserve at the end of the set, whether you’re using 115lbs, 165lbs, or 225lbs. Most importantly for rowing, RPE offers a system of autoregulation based on fatigue from rowing training and other life stressors. RPE8 is still RPE8, even if you did 10 sets of 500 meters the morning of your strength training session, or are under a deadline at work, or it’s exams time at school.

While RPE offers a lot of theoretical advantages, coaches should be aware of the challenges in implementing RPE systems in actuality. Percentage-based strength training has problematic assumptions baked in to the supposedly objective mathematical calculations. RPE-based strength training has problematic assumptions within the assessment framework of implementation.

For example, what do you define as RPE10? Some coaches will say that RPE10 is the point of technical breakdown, where one more rep could not be completed without technical error, such as knee cave on a squat or lumbar spine rounding on a deadlift. Athletes might think that RPE10 is the point of muscular failure, where one more rep is not physically possible regardless of technique. Coaches must choose a definition for each RPE range, and communicate this with athletes. Even if our communication is crystal clear, how do athletes know what RPE10 is unless they experience it? While RPE seemed like a way to avoid rep-max testing, without some work at all the RPE ranges, how can we expect athletes to have accurate reference points?

Another common question is if athletes should be adjusting weight on each set to hit the prescribed RPE range, or if the RPE number only applies on the final set. Without taking huge amounts of rest between each set to allow for full mental and physical recovery, it’s unlikely that the first set of an exercise is an equal RPE to the final set of an exercise using the same weight. Some coaches prefer one, others prefer the other. Some coaches use an RPE range, eg. “RPE 7-8,” to get around this. Perhaps the athlete is around an RPE7 on the first set, but an RPE8 on the final set with the same weight as fatigue sets in. All of this requires programmatic clarity.

Finally, there are the human elements. Rowers are intensely competitive athletes, usually highly intrinsically motivated to improve against one’s own standard of performance, and accustomed to pushing through pain and fatigue. To use an RPE-based system, we must trust that athletes are accurately assessing their readiness to train, and are willing to adjust their weight-selection accordingly. Even if both of those are true, sometimes, we make mistakes. What we thought was RPE8 when we started the set may actually be RPE10 by the time we’re at the final rep.

The very definition of “exertion” is also open to debate, as detailed by Halperin and Emanuel (2019). Gunnar Borg, one of the main RPE developers, defined exertion in his 1998 book as, “the feeling of how heavy and strenuous a physical task is.” Others have termed it, “the subjective intensity of effort, strain, discomfort and/or fatigue that is felt during exercise.” Another takes a more cognitive approach with the definition, “conscious sensation of how hard, heavy, and strenuous a physical task is,” removing the neurophysiological complications of defining “fatigue” and “discomfort.” Still others draw a distinction between “effort” and “exertion,” although these terms are commonly used interchangeably, by defining effort as “the amount of mental or physical energy being given to a task.” Any way you define it, RPE is always going to be subjectively determined. Athletes experience fatigue, discomfort, or strain differently, and train with increased or decreased intensity accordingly. The hope of RPE is that providing a tool for athletes helps them find an intensity for each set and training session that is closer to the ideal than in a percentage-based system.

I found no studies comparing outcomes from percentage-based versus RPE for rowing training, or for other endurance athletes. In general strength training research, Helms et al. (2018) compared the two loading systems in 21 college-age males with strength training experience. Subjects trained three times per week over eight weeks with volume-equated loads ranging from 65-100%1RM or RPE5-10. They found that both groups improved bench press and squat 1RM performance, and muscle thickness of the chest and quadriceps muscles, with no significant difference between groups. There was a slight (non-significant) trend in strength gains for the RPE group.

Graham and Cleather (2019) studied 31 male general trainees, ages 23-33, with at least two years of regular resistance training and demonstrated proficiency in the back squat (130-140kg average) and front squat (111-120kg average). The subjects did a 12-week unsupervised strength training program, with one group using a percentage-based program and the other using RPE. They did a block periodized program with front squat on one day, back squat on another, recording sessions in a logbook and recording RPE throughout the study. Both groups made significant improvements in back and front squat 1RM, with a greater magnitude for the RPE group. The researchers found that the RPE group trained at a higher intensity than the percentage group. They suggest that this was because the RPE subjects could increase training loads more rapidly than the fixed percentage group.

Practical Takeaways for Rowing Training

First, with either method you choose, tracking and recording training sessions in some format is a helpful practice to improve training efficiency, efficacy, and engagement. Both self-coached and coached rowers can benefit from this by at least taking guesswork out of training and recording what weights they used for which exercises and which sets and reps, so that future sessions can build on earlier sessions. We can also go beyond this by adding some context to go along with the pure numbers.

Read More: How to Track Your Rowing Workouts

Overall, I think RPE offers greater advantages and fewer disadvantages than percentage-based training for rowers. Based on the results of the available studies with a general population, we can at least expect RPE strength training to be as effective as percentage-based strength training for gaining strength and muscle. The biggest advantage of RPE for rowing training is offering a system of flexibility around fatigue and readiness to train. Making this work in a coaching context requires intentional planning, clear communication, and consistent presentation with athletes.

Use as clearly defined and transparent of an RPE scale as you can find or produce. Communicate clearly the issue of RPE from first set to final set, and if athletes should be adjusting weight on each set to hit the RPE target. Consider adding additional qualitative details to your rating scale to provide detail on what constitutes a rep or level of challenge. For example, it is common to begin “double-breathing” on squats, or resting at the bottom of dead-stop deadlifts, as fatigue sets in. This introduces intra-set rest, which extends the duration of a set and increases the amount of weight an athlete can use, compared to a more controlled lifting tempo. This artificially dampens RPE values, and RPE researchers typically use timed repetitions or a metronome in their studies to regulate this variable. In the less controlled environment of real-world training, we may find more success by helping athletes be aware of these strategies, and how to use, avoid, or at least account for them in training. As another example, the RPE chart from the Helms et al. (2018) study includes half steps at RPE 7.5, 8.5, and 9.5, to define situations in which the lifter feels that they could make a slight increase in load, but could not complete an additional entire repetition. This level of detail may be useful in your training context.

Emphasize the importance of honest readiness assessment and adjusting weight based on that feedback. Rowers in particular may need help understanding the value of training at submaximal intensities, due to the nature of the sport and the normal social encouragement to push to maximal effort in training. This is another reason that I am generally against using strength tests in any sort of seat-selection or ranking process. It is common in rowing programs to use bench pull, chin-ups, timed circuits, or other strength tests for maximum weight or maximum reps as a factor in ranking rowers. While strength is correlated to rowing performance, there is so much more than goes into a rower’s ability to move a boat. The unintended consequence of strength tests for ranking is creating the environment of more weight, more reps, and more fatigue, when this is often unproductive or actively counterproductive in many training phases. Rowers achieve maximal effort and find plenty of fatigue on the water or on ergometers. Strength training is an opportunity to develop different skills, different movements, and different kinds of effort than yet more “simulation” (which it isn’t, but that’s a rant for another time) of rowing training under high load and fatigue.

Percentage-based programs can work in certain situations, with certain athletes, and with certain strategies in place by coaches. For example, an effective time for a more structured percentage-based program may be the off-season, when interference from hard rowing training is minimal. Athletes with experience 1RM testing may be more suited to percentage-based programming than athletes who lack experience, and will have inaccurate training numbers based on a flawed rep-max test. The major strategies I’ve found effective with percentage-based programs center around making the math element simple. For example, there are numerous technology solutions and apps available for programming and distribution. A simple Excel document and use of basic formulas can work just fine as well, in which athletes enter their 1RM and let the spreadsheet perform the calculations. We can also include “bar math” in the spreadsheet, with what combinations of available barbells and weight plates athletes should use to achieve the weight on the bar. Putting as much of this on autopilot as possible frees up attention and time to do real coaching.

Coaches of rowers should consider the potential problems of percent-to-rep correlations, even in experienced athletes. Rowers are especially well trained at submaximal endurance, and pushing through muscular pain and fatigue. This may disproportionately improve their repetition performance below 80%1RM, as noted in the Desgorces et al. (2010) article, and similar to the endurance runners in the Richens and Cleather (2014) piece. Coaches using percentage-based programs might consider adjusting the percentages to achieve the goal number of reps. For example, if rowers follow the trend of endurance runners and achieve 20 reps at 80%1RM when a “typical” target at 80% would be 10-12 reps, we might use 85% with rowers and see if that results in outputs closer to 10 reps. If the authors of the percent-to-reps correlation charts are happy “guess-timating,” then you should be too!

We also may consider combining methods. For example, if we programmed “3 sets at 80% to RPE9,” the athlete would do a self-determined number of reps at a given percentage. This achieves the general intensity (%1RM) for an amount of reps based on individual characteristics. If an individual athlete’s reps are considerably above or below a target range, we could then decide if we want to adjust the percentage up or down for the next training session.

Some researchers and coaches are experimenting with integrating wearable technology to provide further quantitative and qualitative data beyond RPE or percent alone. Researchers in one study used velocity-measuring PUSH bands with 20 semi-pro male rugby players, and found that the PUSH+RPE group outperformed the RPE-only group in countermovement jump, squat, and bench press gains over six weeks. Here is a blog by the strength coach of the Dutch National Rowing Team on how he uses the GymAware velocity-based training with the rowers he coaches. There are more opportunities to incorporate and study technological integrations like this, though we can expect science to lag behind practice as new technologies are developed.

Ultimately, the goal of percentage-based or RPE rowing training is to create a system of programming so that athletes are training at appropriate intensities for the phase of training and goal of each exercise and session. Being aware of the advantages, disadvantages, and inherent assumptions of each system can help rowers and coaches choose which system best fits their needs. We can implement strategies to mitigate pitfalls and improve training outcomes, so that rowers can keep getting stronger, moving better, and staying healthy thanks to a solid system of strength training for rowing.

Get Rowing Stronger!

“Rowing Stronger: Strength Training to Maximize Rowing Performance” is the comprehensive guide to strength training for rowing, from first practice of the off-season all the way to peak championship race performance, and for everyone from juniors to masters rowers. The second edition is available now in print and e-book.


  1. For many years my training has been based on Matt Brzycki’s “one-set-to-failure” prescription. This works well for tracking progress – with constant weight, reps to failure is a measure of progress. I increase the weight when I progress to a limit number of reps. This measure is precise with the leg press (sled-type), because failure is uncontrollable shakes that can’t be “worked through.” I am a rower and use less weight & more reps than Brzycki recommends – when queried, he said, “That makes it more of an endurance exercise.” He’s right, and that’s the idea.

    1. Hi Carlo,

      Sure. Training to total failure on machines certainly does make recording workout data simple. This is not the kind of training I use with competitive rowers, so some method of navigating weight selection for submaximal training loads becomes a necessity.


Leave a Reply

Your email address will not be published. Required fields are marked *