Member Portal Area

Pushing It to the Limit: Gauging How Far We Are From Failure

Mass Research (Volume 1 - Issue 1)

Study Reviewed: Accuracy in estimating repetitions to failure during resistance exercise. Hackett et al. (2016)

Key Points

  1. Both trained and untrained men and women can accurately (within 1 repetition) gauge how many more repetitions they can perform when they are 0-5 repetitions from failure when performing machine chest press.
  2. Trained and untrained men can also gauge repetitions remaining, with the same degree of accuracy, when 0-5 repetitions from failure on the leg press. However, trained and untrained women become inaccurate when they have more than 3 repetitions remaining before failure.
  3. Based on prior research and the low (yet non-significant) p-value associated with training age, lifters with greater training experience may be able to more accurately gauge repetitions remaining before reaching failure.
  4. When greater than 5 repetitions (for men on the leg press or men and women on the chest press) or 3 repetitions (for women on the leg press) remain, lifters systematically under-predict how many more repetitions they can perform before reaching failure.

The researchers set out to measure the accuracy of estimating how many repetitions in reserve (RIR) can be performed before reaching failure. They also set out to determine if accuracy of estimating RIR was influenced by training experience, exercise performed, or sex. Participants performed multiple sets (up to 10) with 70% 1RM on chest press and 80% 1RM on leg press. After completing 10 repetitions, participants verbally stated RIR, then immediately continued performing repetitions until failure to compare estimated versus actual RIR. When combining outcomes of both lifts, mean estimated RIR was off by about 1 from actual RIR, when there were 0 to 5 actual RIR remaining. However, when actual RIR was 7 to 10, estimated RIR were off by more than 2. Comparing chest press to leg press, the average error in estimated RIR was less than 1 repetition for chest press when actual RIR was 0 to 5. However, for the leg press, this was only true when actual RIR was 0 to 3. Males were more accurate at assessing RIR than females during leg press only. Training experience did not impact accuracy. Since lifters can accurately estimate RIR, it can be used for prescribing and tracking intensity.

Purpose and Research Questions

These researchers previously conducted a similar experiment on a smaller group of male bodybuilders performing the back squat and bench press (1). However, in the present study, they set out to determine the accuracy in a larger, mixed-sex group with varied levels of training experience performing machine leg and chest press. They hypothesized RIR accuracy would be greater when closer to failure. Additionally, they hypothesized accuracy would be greater among more experienced lifters, and would not be influenced by sex or exercise type.Subjects and Methods

53 males and 28 females with varying resistance training experience participated. 16 had ≤ 6 months of training experience, 14 had 1 to 2 years, and 51 had ≥ 3 years.

The participants performed 1RM tests, and 48 hours later performed multiple sets of 10 repetitions with 70% and 80% of 1RM on the chest press and leg press, respectively. These 1RM percentages were chosen as they resulted in performance of ≤ 20 repetitions in pilot testing.

After completion of each set, participants stated how many more RIR they believed they could perform. Participants continued performing sets (to a maximum of 10) until they couldn’t complete at least 10 repetitions in a set. Findings

Actual RIR after completing the prescribed 10 repetitions on both chest and leg press – for men and women – ranged from 0 to 10. Combined leg and chest press
analysis revealed there was greater accuracy estimating RIR the closer to failure one was. Additionally, participants more accurately assessed RIR with chest press compared to leg press, and males were more accurate than females. Finally, resistance training experience did not impact accuracy (the p-value was low, but not quite down to 0.05).

For chest press specifically, estimated RIR accuracy was on average ≤ 1 repetition from actual RIR. However, this level of accuracy was only obtained when actual RIR was 0 to 5. As actual RIR increased, estimation accuracy decreased. The same trend was found for leg press. However, accuracy fell off earlier for leg press, with an error of ≤ 1 repetition only when actual RIR was 0 to 3.

Both lifts followed the same pattern in that when participants were near failure (0-3 actual RIR for leg press and 0-5 for chest press), they either accurately estimated RIR or just slightly overestimated (on average by ~1 repetition). However, when they were at more than 5 actual RIR for chest press and more than 3 for leg press, participants systematically underestimated RIR. Furthermore, the degree of underestimation increased when further from failure. For example, when there were 6 actual RIR, participants on average estimated 4 or 5 RIR, underestimating by 1-2. However, when there were 10 actual RIR, participants on average estimated 5 to 7 RIR, underestimating by 3-5.

The difference in accuracy between sexes was relegated to leg press, and furthermore, only existed when estimating RIR if there were 4 or more actual RIR. Meaning, when there were 0 to 3 actual RIR on the leg press, males and females estimated RIR with the same accuracy.

Adapted from Zourdos et al. (2)


This study is the most direct measurement of RIR accuracy to date. While Dr. Zourdos and I validated an RIR-based RPE scale with velocity (see figure above), showing higher RPE scores (indicated by fewer RIR) correspond to decreases in bar velocity (2, 3), we did not directly measure actual versus estimated RIR. Hackett and colleagues were the first to do so in 2012 (1). However, their 2012 study primarily presented accuracy with correlations between actual and estimated RIR. This 2016 study presents data in a clearer way, showing where breakpoints in accuracy occur and the degrees of under- and overestimation.

In contrast to the recent study by Zourdos and colleagues, which reported novice lifters as less accurate than experienced (2), Hackett found RIR accuracy was not influenced by training experience. However, the p-value of the difference between novices and experienced lifters was 0.134.

P-values are statistics that estimate the likelihood findings are due to chance. Meaning, the difference in accuracy between inexperienced and experienced lifters only had a 13.4% chance of being random statistical variation.

Thus, while Hackett – in a strict sense – found no significant difference between novice and experienced lifters, there may have been one. Furthermore, the least experienced participants in the present study had ≤ 6 months of experience, while the most experienced had ≥ 3 years of experience. In Zourdos’ investigation, there was a greater gap with a mean of 0.4 years of experience in the novice group, and an average of 5.2 years among the experienced group, which may have been a large enough gap to show a difference. Finally, this discrepancy may have been due to using different exercises. Zourdos’ participants performed the back squat, while Hackett’s performed machine leg press, which arguably takes less experience to master.

In contrast to another recent study on RIR-based RPE differences in powerlifters that Dr. Zourdos and I published, Hackett and colleagues found greater accuracy in the chest press compared to the leg press. However, we found no differences in RPE among the squat, bench press, or deadlift (3).

Hackett et al. proposed that greater accuracy in the chest press may be due to a higher sensory organ density in the upper versus the lower limbs (4). They also point out that this may be the reason women were less accurate only in the leg press, as males have a higher sensory organ density in the lower body (4). While plausible, it is also possible that the greater perceived effort of going to failure during lower body versus upper body exercises (5) resulted in participants having less prior experience going to failure on leg press than chest press. To explain the sex difference, perhaps societal differences in male versus female lifting culture (the “go hard or go home” mentality) encourage men to push themselves to failure more regularly on leg press, thus better developing their awareness of RIR. This may be why we found no difference in RIR-based RPE among powerlifters performing upper and lower body exercises, while Hackett did.

Interestingly, the authors address the utility of the RIR-based RPE scale that Dr. Zourdos and I study (see figure on previous page), which was originally developed by powerlifter and coach Mike Tuchscherer in 2008 (6). They state this scale is not intuitive, as RPE and RIR are inversely related, adding an unnecessary layer of complexity by forcing the user to translate RIR into an RPE. I don’t completely disagree with this notion. However, this is easily alleviated by allowing participants to report whichever they are most comfortable with, either RPE or RIR. Then, the investigator confirms the opposite to ensure understanding. For example, when a participant completes a set, we show them the scale and they are free to state “I had 2 repetitions remaining.” At that point we confirm “so an 8 RPE?,” or vice versa.

Additionally, the RIR-based RPE scale has some advantages over reporting raw RIR. In the initial study introducing the scale, novice and experienced lifters worked up to a 1RM, using the scale to determine when it was achieved. Specifically, if an RPE less than 10 was called, an increase in load was made on the subsequent set. If the attempt was missed, the participant was “stuck” with an RPE less than 10. Novice lifters were “stuck” with an average RPE of 8.96, while experienced lifters achieved a 9.8 (2). An average RPE of 9.8 was due to a large proportion of lifters achieving 9.5 RPE scores in attempts leading up to their final, then making small jumps to try to reach a 10 RPE. Thus, it seems experienced lifters can tell with more accuracy than novices when they are less than 1 RIR from failure. This shows the value in having a scale which allows 9.5, 8.5, and 7.5 values, especially considering that the data from the current study by Hackett show similar accuracy when 0-3 RIR remain.

In fact, the structure of the RIR-based RPE scale is supported by Hackett’s findings. Half-point RPE scores are not used below an RPE 7 (RIR 3). Rather, RPE 5-6 represents 4-6 RIR, while RPE values below 5 are given subjective terminology indicating effort. This aligns well with the present findings, considering accurate estimations are difficult when greater than 3 RIR remain. With that said, RPE scores below 7 still have utility. As we proposed in a recent review article, when explosive lifting far from failure is the goal (i.e. velocity dominant power training), if the lifter feels they can accurately gauge RIR, the load is too high or too many repetitions are being performed with that load. Thus, an “RPE cap” can be used for power training whereby 1-6 RPE values can be used depending on whether force (5-7 RPE) or velocity (1-4 RPE) dominant power training is the goal (7).

With all that said, lifters should feel free to use either the RIR-based RPE scale above or raw RIR, based on whichever is more intuitive. You can avoid the downside of raw RIR not having half-point values by simply recording <1 RIR to be equivalent to a 9.5 RPE, and 1-2 and 2-3 RIR to be equivalent to an 8.5 and 7.5 RPE, respectively. Next Steps

At this point, thanks to Hackett and colleagues, the accuracy of RIR as a measure of resistance training effort has been well established. The next step is to conduct training studies in which resistance training intensity is prescribed with an RIR or RIR-based RPE value and compare that to a traditional model of training using percentage 1RM to see if there is any advantage or disadvantage to doing so. Furthermore, based on my anecdotal coaching experiences, certain personalities are better suited to gauging RIR than others. Thus, future research should strive to determine how to identify which individuals are suited to using this method of gauging effort in training and which individuals are not.

Application and Takeaways

  1. Stress per set can be accurately gauged by most lifters using RIR or RPE-based RIR when 0-3 repetitions from failure. When using pure RIR, the accuracy of values greater than 5 should be viewed with caution. Values greater than 3 should be viewed with caution when used by women performing lower body movements. However, this may not be the case in well-trained female athletes.
  2. Compared to pure RIR, the RPE scale based on RIR has the advantage of using broader repetition ranges and more subjective language after 3 RIR to account for decreasing accuracy further from failure. However, it also has an extra layer of interpretation as RIR must be translated to RPE. Thus, RIR can be recorded and then converted to RPE later if so desired.
  3. Given prior research on the accuracy of RPE when used by novices compared to experienced lifters and the low p-value associated with training age reported in this study, lifters new to gauging RIR or RPE should go through a familiarization period. Thus, a mesocycle in which load is programmed with percentage of 1RM and RIR or RPE is simply recorded, should occur before a mesocycle where load is prescribed with RIR or RPE values.


  1. Hackett DA, Johnson NA, Halaki M, and Chow CM. A novel scale to assess resistance-exercise effort. J Sports Sci 30: 1405-1413, 2012.
  2. Zourdos MC, Klemp A, Dolan C, Quiles JM, Schau KA, Jo E, Helms E, Esgro B, Duncan S, Garcia Merino S, and Blanco R. Novel Resistance Training-Specific Rating of Perceived Exertion Scale Measuring Repetitions in Reserve. Journal of strength and conditioning research 30: 267-275, 2016.
  3. Helms ER, Storey A, Cross MR, Brown SR, Lenetsky S, Ramsay H, Dillen C, and Zourdos MC. RPE and Velocity Relationships for the Back Squat, Bench Press, and Deadlift in Powerlifters. Journal of strength and conditioning research, 2016.
  4. Han J, Park S, Jung S, Choi Y, and Song H. Comparisons of changes in the two-point discrimination test following muscle fatigue in healthy adults. Journal of physical therapy science 27: 551-554, 2015.
  5. Shimano T, Kraemer WJ, Spiering BA, Volek JS, Hatfield DL, Silvestre R, Vingren JL, Fragala MS, Maresh CM, Fleck SJ, Newton RU, Spreuwenberg LP, and Hakkinen K. Relationship between the number of repetitions and selected percentages of one repetition maximum in free weight exercises in trained and untrained men. Journal of strength and conditioning research 20: 819-823, 2006.
  6. Tuchscherer M. The Reactive Training Manual: Developing your own custom training program for powerlifting. Reactive Training Systems, 2008, p 15.
  7. Helms ER, Cronin J, Storey A, and Zourdos MC. Application of the Repetitions in Reserve-Based Rating of Perceived Exertion Scale for Resistance Training. Strength and conditioning journal 38: 42-49, 2016.

Go back to articles
Don't be antisocial! - Join our exclusive groups here!