AMRAP Sets to Failure vs. Repetitions in Reserve for Strength

Terminating sets with 1 to 2 repetitions in reserve (RIR) produces maximal strength adaptations equivalent to—and often slightly superior to—taking sets to technical failure, while inducing substantially less acute and residual fatigue [2, 3, 10, 15]. However, taking open-ended AMRAP ("as many repetitions as possible") sets to technical failure serves as a highly effective autoregulatory tool to objectively calibrate subsequent training loads [18, 19, 25]. The primary trade-off over a training cycle is that while failure sets accurately gauge progression, they cause pronounced acute velocity loss, elevated biochemical stress markers, and delayed recovery [10, 16].

Strength Adaptations: Proximity to Failure

Muscular strength adaptations are primarily driven by high mechanical tension, motor unit recruitment, and neural adaptations such as heightened motor unit synchronization and reduced antagonist co-contraction [4, 6]. Because high-load resistance training provides superior neural drive adaptations compared to lighter loads [6], lifting heavy loads does not require absolute failure to achieve complete motor unit recruitment [2, 15].

A comprehensive meta-regression by Robinson and colleagues analyzing 67 strength studies established that training closer to failure did not benefit 1RM strength gains; in fact, closer proximity to failure was associated with slightly smaller strength increases due to excessive fatigue interference [2]. Continuous multi-level modeling revealed that the marginal slope for proximity to failure and strength development had a confidence interval containing zero, confirming a negligible relationship between training to failure and strength progression [3]. These findings align with earlier meta-analytic evidence by Davies and colleagues, which demonstrated that resistance training to failure yielded strength gains statistically indistinguishable from non-failure training [2].

Direct randomized controlled trials reinforce these meta-analytic outcomes. In a 10-week investigation of resistance-trained men, training to failure and non-failure protocols led to statistically similar improvements in leg press 1RM (22.3% vs. 26.7%) and leg extension 1RM (33.3% vs. 33.7%), alongside identical electromyographic (EMG) amplitudes and muscle architectural changes [15]. Similarly, an 8-week trial by Refalo and colleagues comparing unilateral leg training to momentary failure versus stopping at 1 to 2 RIR found no measurable performance or growth advantage for the failure condition [2, 11].

Fatigue Mechanisms and Recovery Kinetics

The physiological cost of reaching failure is substantial. During resistance exercise, intramuscular disturbances—including the depletion of phosphocreatine (PCr), adenosine triphosphate (ATP), and glycogen, combined with the accumulation of hydrogen ions ($H^{+}$) and inorganic phosphate—trigger peripheral fatigue and impair contractile function [4]. These metabolites activate mechanosensitive group III and metabosensitive group IV unmyelinated muscle afferents, which project inhibitory feedback to the central nervous system, blunting central motor drive [4].

Acute neuromuscular fatigue scales directly with proximity to failure. When resistance-trained individuals completed 6 sets of bench press at 75% of 1RM, acute velocity loss measured 4 minutes post-exercise was 25% following failure sets, compared to 13% at 1 RIR and 8% at 3 RIR [10]. This velocity deficit is modulated by biological sex, with males experiencing greater acute velocity loss (-29%) than females (-21%) at 4 minutes post-failure [10].

Residual fatigue also follows distinct timelines based on proximity to failure:

• 24-Hour Recovery: Velocity decrements remain depressed at 24 hours post-exercise following both failure (-3%) and 1 RIR (-3%) conditions, whereas leaving 3 RIR allows velocity to recover (+2%) [10]. Protocols taken to failure also cause marked elevations in serum creatine kinase, ammonia, and acute reductions in countermovement jump height compared to non-failure schemes [4, 16].
• 48-Hour Recovery: Between-protocol velocity differences largely resolve by 48 hours [10]. However, consecutive training sessions without 48 hours of recovery lead to persistent elevations in serum creatine kinase, elevated salivary cortisol, and suppressed salivary immunoglobulin A (IgA) [17].

AMRAP Sets as an Autoregulatory Tool

While taking every set to failure generates unnecessary fatigue, open-ended AMRAP sets (also known as "plus sets") provide an objective solution to the problem of individual progression rates [19, 23]. Human strength responses to identical resistance training programs are highly variable, with 1RM improvements ranging from 0% to 250% across individuals over 12 weeks [19, 23]. Fixed, arbitrary weekly progressions (such as adding 2.5 kg to 5 kg every week) fail to accommodate these disparate adaptation rates [19].

Autoregulating Progressive Resistance Exercise (APRE) leverages a final AMRAP set taken to failure to adjust subsequent loading [19, 25]. In standard protocols (such as APRE3 for maximal strength or APRE6 for strength and hypertrophy), the repetitions completed on the test set determine the exact load adjustment for the next set or training week [19, 25]. A systematic review and network meta-analysis found that APRE ranked highest among autoregulation strategies for maximal strength development, achieving surface under the cumulative ranking curve (SUCRA) scores of 93.0% for the back squat and 97.1% for the bench press [18]. For bench press 1RM, percentage-based progression demonstrated a large negative effect compared to APRE (Standardized Mean Difference [SMD] = -0.83), while subjective RPE-based prescription showed a moderate negative effect relative to APRE (SMD = -0.76) [18].

To avoid disproportionate progression increments between lifters of differing strength levels, APRE load adjustments are more balanced when scaled as percentages of 1RM (+1% to +3.5%) rather than fixed absolute weights [19].

Subjective RIR Gauging and Cycle Management

For lifters seeking to manage fatigue without using AMRAP sets to failure on every exercise, prescribing loads via subjective RIR is an effective alternative [23]. Subjective RIR estimation improves significantly with training experience [7]. Experienced lifters demonstrate a strong inverse correlation between movement velocity and subjective RIR ratings ($r=-0.88$) and can accurately rate 1RM attempts compared to novices [7]. In bodybuilding populations, repetition predictions at 70% 1RM correlate highly with actual failure performance in the squat ($r=0.93$) and bench press ($r=0.95$), with estimation precision increasing as the set nears failure [7].

Across trained populations, lifters typically estimate 1 RIR targets within 0.40 repetitions of true failure [11]. In lower repetition ranges (3 to 5 repetitions), RIR reliability is particularly robust, displaying lower coefficients of variation (~3–4%) than higher rep ranges (~6%) [8].

Practical Application Over a Mesocycle

Balancing progression and fatigue management over a multi-week training cycle involves distinct roles for both strategies:

1. Primary Volume Sets (1–2 RIR): The majority of working sets across compound movements are best maintained at 1 to 2 RIR. This threshold provides optimal mechanical tension and muscle activation while preventing excessive group III/IV afferent-mediated central fatigue and persistent 24- to 48-hour neuromuscular deficits [2, 4, 10, 15].
2. Strategic AMRAP Plus Sets: Positioning a single open-ended AMRAP set on the final set of a primary exercise provides an objective performance marker to dictate load adjustments for the subsequent week [19, 25].
3. Recovery Spacing: When AMRAP sets to technical failure are utilized, programming at least 48 hours of recovery before retraining the same muscle groups ensures normalisation of resting neuromuscular velocity and biochemical recovery [10, 16, 17].