Accentuated eccentric loading (AEL) uses the greater load lifters can resist during lowering by applying more load on the way down than on the way up. Weight releasers, computer-controlled resistance, pneumatic devices and carefully coordinated manual methods can all create that phase-specific difference.
Bench studies report greater eccentric force, work and sometimes muscle activation with AEL. A 2026 meta-analysis found slightly lower concentric velocity on the exposed repetition and similar velocity on the next two. A separate 49-study synthesis found greater acute activation, metabolic stress and effort, with comparable long-term strength and muscle-size gains to constant loading.
What changes when the lowering load is heavier
- AEL is a loading definition, not a device. The eccentric load must exceed the paired concentric load.
- The eccentric stimulus increases. Bench experiments report more lowering-phase force, work and pectoralis activity under some protocols.
- Immediate concentric speed is similar or slightly slower. Pooled evidence shows a small first-repetition velocity decrement and similar later repetitions.
- Long-term adaptations are comparable on average. The newest broad synthesis found similar strength and cross-sectional-area outcomes versus constant loading.
- Implementation is part of the intervention. Release timing, stabilization, range, spotters, familiarization and the load differential can decide whether the set is trainable.
Record the eccentric and concentric loads together, followed by the device, repetitions, range of motion and release mechanics. These details specify a reproducible AEL exposure.
What counts as accentuated eccentric loading?
A slow eccentric with the same bar weight is eccentric-emphasis training, but it is not AEL. Neither is a flywheel repetition automatically AEL; eccentric overload must actually occur and be quantified. In AEL, the external load during lowering is greater than the load during the subsequent press, and reviews emphasize that the eccentric and concentric components must be reported as a pair [1][2]. Zhang and colleagues classified protocols relative to concentric 1RM as submaximal (eccentric load below 1RM), maximal (equal to 1RM) or supramaximal (above 1RM) [3].
Four ways to deliver the eccentric load
Weight releasers

Plate-loaded releasers extend from the bar to the floor. Look for the long hooks and extra plates outside the bench.
Photo source: Stray Dog Strength
Hooks carry extra plates down and detach near the bottom. They standardize load but usually apply overload only to selected repetitions and can disturb the turn-around.
Computer-controlled resistance

Vitruvian Trainer+ in use. Cables connect the bar to the motorized platform beneath the bench.
Photo source: Vitruvian
Motorized or cable systems can change resistance by phase on every repetition. The device’s load definition, calibration and motion constraints must be reported.
Pneumatic systems

Keiser seated chest press in use. Air cylinders and handle-mounted load controls identify the pneumatic system.
Photo source: Impact Your Fitness
Pneumatic cylinders use compressed air. With suitable controls, resistance or assistance can change between phases and reduce plate handling. Phase-specific loading depends on the device and protocol; its settings are not automatically interchangeable with free-weight percentages.
Manual loading

Manual eccentric-loading demonstration. Two spotters work at the bar ends to assist the lift and apply resistance during lowering.
Photo source: Pete Burridge / SimpliFaster
Spotters add or remove force or plates. This is difficult to quantify and coordinate; it needs more personnel and carries greater execution uncertainty.
Write a prescription someone else can repeat
Load pair
Lowering load / pressing load, with the same stated reference.
Repetition exposure
Every rep or first rep; tempo, pause and range.
Session structure
Set count, rest and the transition between reps.
Write both loads using the same reference: 100/70 means 100% 1RM down and 70% up. Include which reps receive overload, tempo, pause or rebound rule, ROM, transitions, rest and set count.
Practice came before clean comparisons
Josh Bryant and Louie Simmons described weight-releaser use in powerlifting before modern pooled analyses.[4][5] Their accounts document how coaches implemented the method; the studies below compare its effects.
One repetition. Two loading phases.
Lower with the added load
The releasers travel down with the bar.
Release at the planned transition
The added load detaches before the press.
Press the lighter bar
The remaining bar load sets the concentric task.
Illustrative weight-releaser sequence. Match setup and release procedure to the intended protocol.
Acute bench evidence: more eccentric work, inconsistent concentric benefit
Direct mechanical studies
Castro, Zangakis and Moir studied 12 resistance-trained men with an average bench 1RM of 134 ± 33 kg. In a repeated-measures design, each performed two-repetition sets under four conditions: traditional 30% and 80% 1RM, or the same concentric loads preceded by a 100% 1RM eccentric using weight releasers. Force plates, three-dimensional motion capture and surface EMG quantified the lift.
What changed during the lowering phase?
12 trained men · two-repetition sets. Values below are measured lowering-phase outcomes, grouped by the matched pressing load.
30% pressing condition
Traditional: 30% down / 30% up
AEL: 100% down / 30% up
80% pressing condition
Traditional: 80% down / 80% up
AEL: 100% down / 80% up
Lowering force
N · mean ± SD
Traditional 30/30; AEL 100/30 (% 1RM down/up). [6]
Lowering work
J · mean ± SD
Traditional 30/30; AEL 100/30 (% 1RM down/up). [6]
Pectoralis sEMG
μV · mean ± SD
Traditional 30/30; AEL 100/30 (% 1RM down/up). [6]
Lowering force
N · mean ± SD
Traditional 80/80; AEL 100/80 (% 1RM down/up). [6]
Lowering work
J · mean ± SD
Traditional 80/80; AEL 100/80 (% 1RM down/up). [6]
Pectoralis sEMG
μV · mean ± SD
Traditional 80/80; AEL 100/80 (% 1RM down/up). [6]
Bars show Table 2–3 condition means; labels include SD. Each metric uses the same zero-based scale across the two pressing conditions. These are descriptive condition comparisons: the paper’s overall main effects should not be read as separate p values for every pair. Force, work and sEMG have different units. [6]
The study reported overall AEL-versus-traditional mean differences of 118 N in force, 43 J in work and 27 μV in pectoralis activity; all were statistically significant. These overall effects combine lifting phases and loading conditions. The condition-specific lowering means are shown above. There was little evidence that those changes potentiated the following concentric phase [6].
Track where the repetition changes
Tested load pair
Kristiansen, Larsen and van den Tillaar tested a still heavier pair in 10 resistance-trained men: two reps at 110% down/85% up versus 85/85. Three-dimensional kinematics and EMG were analyzed through pre-sticking, sticking and post-sticking regions.
Observed second repetition
On the second AEL repetition, the sticking region began lower and bar velocity fell, while some pectoralis and deltoid EMG measures increased. The authors interpreted this pattern as fatigue.[7]
The load pair changes the question
Other acute trials show why results remain specific to the load pair, outcome and athlete.
Doan et al.
105% lowering loadDoan and colleagues reported that eight trained participants lifted roughly 3% more after lowering 105% 1RM with detachable hooks [8].
Ojasto & Häkkinen
Load-dependent responseOjasto and Häkkinen, by contrast, found that 120% eccentric loading reduced the maximal concentric load from 108.4 ± 3.6 to 104.5 ± 3.3 kg; a lighter individualized eccentric condition paired with 50% concentric loading improved power [9].
Van den Tillaar & Kwan
95/85 vs. 85/85Van den Tillaar and Kwan found no kinematic advantage for 95/85 versus 85/85 across three reps in 16 trained men [10].
The 2026 repetition-by-repetition synthesis
Yang and colleagues pooled eight acute crossover studies with 106 participants: 36 effects for the AEL repetition, 21 for the second and 22 for the third. Most participants met entry criteria of at least a bodyweight bench or 1.5-times-bodyweight squat.[11]
AEL slightly reduced first-repetition mean concentric velocity: effect size −0.25 (95% CI −0.33 to −0.16; p < .001). The bench subgroup estimate was −0.11; the squat estimate was −0.33.
The second repetition was essentially identical to traditional loading (0.01; 95% CI −0.10 to 0.12), and the third showed a trivial, non-significant estimate (0.08; 95% CI −0.08 to 0.23). Evidence certainty fell from moderate for the first rep to low and very low for the second and third.
Read the repetition and the uncertainty together
Mean concentric velocity across three repetitions
After the added load is released
Standardized effect · estimate [95% confidence interval]
Points show estimates; lines show 95% confidence intervals. The vertical line marks zero. [11]
Negative values indicate slower velocity with AEL; positive values indicate faster velocity. The pooled bench and squat studies applied overload to the first lowering phase only.[11]
The acute experiment must respect the load pair
In 18 strength-trained men, Michalak and colleagues compared traditional loading with AEL conditions using eccentric loads equal to 100% or 110% of 1RM across concentric loads from 40% to 80%. Concentric load magnitude explained more of the measured performance than condition, while AEL altered mechanics without producing a consistent performance benefit [12]. The load pair is therefore part of every AEL result.
Name both loads and their reference
Lowering phase ↓
1.5×Relative to the pressing load
Pressing phase ↑
1.0×The reference load
Normalized illustration of a 150%-of-concentric-load relationship. It is an explanation of the load pair, not a recommended protocol.
Article synthesis; see the references below.
Choose the outcome before the set
Did eccentric work rise?
Use phase-specific force, impulse or work.
Did the press improve?
Use matched concentric velocity, load or power.
Did training adapt?
Use a longitudinal constant-load comparator.
Chronic adaptation: a distinct question
Zhang and colleagues searched four databases through July 2024 and included 49 studies with 773 participants. Acute and chronic outcomes were pooled separately, and chronic interventions had to last at least four weeks.
Compared with constant-load training, AEL increased acute eccentric-phase EMG (SMD 0.37, p = .01), lactate (0.44, p = .03), growth hormone (0.50, p = .01) and perceived exertion (1.72, p = .01).
For longitudinal outcomes, the pooled adaptations were comparable: maximal concentric strength SMD 0.12 (p = .41), maximal eccentric strength 0.19 (p = .58), isometric strength 0.03 (p = .93), countermovement jump 0.04 (p = .87), and muscle cross-sectional area −0.06 (p = .84). Fascicle-length uncertainty was wide (0.90, p = .17).
Across the included exercises and protocols, average training adaptations were comparable.[3]
Immediate response and adaptation, side by side
During or immediately after exercise
Standardized effect · estimate [95% confidence interval]
Points show estimates; lines show 95% confidence intervals. The vertical line marks zero. [3]
After a training block
Standardized effect · estimate [95% confidence interval]
Points show estimates; lines show 95% confidence intervals. The vertical line marks zero. [3]
Both panels use the same standardized-effect scale; positive values favor AEL. All displayed chronic confidence intervals cross zero. In a sensitivity analysis of matched prescriptions, the lactate difference was not significant.[3]
What the pooled estimates mean for programming
How to read a pooled estimate
Average effect
A standardized effect is not a kilogram change.
Uncertainty
An interval spanning zero leaves both directions compatible with the estimate.
Protocol context
Device, load ratio and training status still matter.
The pooled differences in chronic concentric strength (0.12, p = .41) and muscle cross-sectional area (−0.06, p = .84) were small and nonsignificant across the studied exercises, devices and loading protocols.
More eccentric work can coexist with a slower press
Lowering phase
Force, work and muscle activity describe the eccentric exposure.
Pressing phase
Velocity and sticking-region mechanics describe the subsequent effort.
Castro and colleagues found greater eccentric force, work and pectoralis activity with 100% lowering loads paired with 30% or 80% pressing loads. Yang’s pooled analysis found slightly slower pressing on the exposed repetition. Kristiansen’s 110/85 protocol also changed sticking-region mechanics. Higher eccentric exposure can therefore come with a slower immediate press.
Match the evidence to the goal
The programming implication is to judge AEL against its declared purpose.
Phase-specific exposure
If the purpose is phase-specific exposure, audit eccentric load, descent control, work and tolerance while accepting that ordinary concentric speed may not rise.
Acute potentiation
For acute concentric potentiation, compare press velocity with an otherwise matched traditional set and retain the method only when the athlete responds as intended.
Training adaptation
If the purpose is long-term strength or hypertrophy, use a training block with preplanned outcome tests and a conventional comparison where feasible.
Choose the success metric before the set and keep it consistent. A “faster press” and a “harder eccentric” are different objectives.
When AEL is a rational choice
Readiness starts with a repeatable task
Clear reason
Know why the eccentric phase needs extra loading.
Controlled transition
Match equipment, staff and practice to the release.
Repeatable setup
Establish the bottom position under ordinary loading first.
AEL earns a place when the coach wants a higher absolute eccentric exposure while preserving a pressable concentric load, has the equipment and personnel to control the transition, and can explain why ordinary loading is insufficient. It may fit a short strength block, phase-specific familiarization or a research-informed experiment in a well-trained lifter. It is a poor default for novices, for unsupervised maximal work, or when the athlete cannot reproduce the bottom position under ordinary loading.
Stronger lifters appeared to tolerate AEL better in some studies, although Yang and colleagues could not pool this relationship. Assess technical readiness directly through supervised familiarization and control of the release.
Programming Eccentric Overload for the Athlete and Goal
A decision card for each part of the session
Experience
- Conservative entry
- Stable full-range technique and several supervised familiarization exposures
- Progress only when
- The athlete controls the descent and release without altered touch point
- Stop or regress when
- Bracing, bar path or contact points change
Load pair
- Conservative entry
- Begin submaximal or maximal eccentrically, with a clearly pressable concentric load
- Progress only when
- Velocity, RPE and technique remain within plan
- Stop or regress when
- The first press slows unexpectedly or assistance is needed
Volume
- Conservative entry
- Low rep counts and few AEL exposures; ordinary work supplies most weekly volume
- Progress only when
- Recovery and next-session performance are stable
- Stop or regress when
- Soreness, perceived effort or performance cost exceeds the block’s purpose
Rest
- Conservative entry
- Longer than ordinary power work; reset releasers deliberately
- Progress only when
- Each attempt begins from a repeatable setup
- Stop or regress when
- Handling pressure creates rushed transitions
Staffing
- Conservative entry
- Competent handoff plus side spotters and correctly set safeties
- Progress only when
- Every person has rehearsed the release and abort procedure
- Stop or regress when
- Staff or equipment cannot reproduce the protocol
Name the success measure
- Before the session
- Specify eccentric exposure, press performance or a training-block outcome.
- At review
- Compare the result with the preselected goal.
- Keep the question consistent
- A harder descent does not by itself satisfy a faster-press target.
An entry block, written as a complete task
2–4 sets
One to three repetitions per AEL set.
1–2 sessions / week
Most bench volume remains conventional.
Prewritten stop rules
Record the load pair, ROM, rest and velocity or RPE limits.
As a coaching starting point, consider two to four AEL sets of one to three reps after the normal warm-up, once or twice weekly, while keeping most bench volume conventional. Record both loads, ROM, rest and velocity or RPE stopping rules. When speed is the goal, compare the press against a matched traditional set.
Rehearse the transition before adding exposure
Weight releasers must clear the floor and bench consistently without striking the athlete or frame. Set safeties for the concentric load and the athlete’s actual bottom position; do not assume the hooks will detach perfectly. Preserve the external contact conditions described in Spinal Positioning & Force Transfer. Manual methods require more, not less, risk planning. Supramaximal exposures should never be improvised because a previous set “felt easy.”
The four-week Launch Pad study in the wider literature
Goldman and colleagues randomized 42 intermediate-trained men to conventional flat-bench or Launch Pad training for four weeks. Both groups trained three times weekly for 12 sessions on a computer-controlled resistance device, performing five sets to failure with 150% eccentric relative to concentric resistance. The comparison therefore held the AEL method in both groups while changing the athlete-to-bench interface.
Strength gains under the same AEL program
A shared four-week program
- Participants
- 42 intermediate-trained men
- Exposure
- 12 sessions · three sessions per week
- Held in common
- Computer-controlled AEL, five sets to failure, eccentric resistance at 150% of concentric
- Compared
- Conventional flat-bench versus Launch Pad interface
Four-week 1RM gain
kg gained · mean ± SD
7.3 kg between-group difference · p < .001 · Hedges g = 3.85. [13]
Bars show reported mean changes; labels include SD. The shared zero-based scale shows gains in kilograms. No error bars are drawn. This study compares interfaces within an AEL program. [13]
Conventional bench 1RM increased 11.1 ± 2.4 kg; the Launch Pad group increased 18.4 ± 4.3 kg. The between-group difference was 7.3 kg (p < .001; Hedges g = 3.85) [13].
What this comparison contributes
Tested comparison
Both groups completed the same AEL program; the Launch Pad® group gained 7.3 kg more 1RM strength on average.[13]
Treat AEL as a measured exposure
Start with the target outcome, record both phase loads, and let concentric performance and technical fidelity decide whether the method stays in the program.
Safety: use appropriate equipment, competent spotters, safeties and supervised familiarization for AEL. Do not improvise unsupervised supramaximal loading.
References
- Wagle, J. P., Taber, C. B., Cunanan, A. J., Bingham, G. E., Carroll, K. M., DeWeese, B. H., et al. (2017). Accentuated eccentric loading for training and performance: a review. Sports Medicine, 47, 2473–2495. https://doi.org/10.1007/s40279-017-0755-6 ↩
- Taber, C. B., Morris, J. R., Wagle, J. P., & Merrigan, J. J. (2021). Accentuated eccentric loading in the bench press: considerations for eccentric and concentric loading. Sports, 9, 54. https://doi.org/10.3390/sports9050054 ↩
- Zhang, X., Weakley, J., Li, H., Marcos-Frutos, D., & García-Ramos, A. (2026). Acute and chronic effects of accentuated eccentric loading vs. constant-load resistance training: a systematic review and meta-analysis. Sports Medicine, 56, 1749–1770. https://doi.org/10.1007/s40279-026-02422-7 ↩
- Bryant, J. (2013). Bench Press: The Science (pp. 75–76). JoshStrength LLC. Practitioner source supplied for historical context. ↩
- Simmons, L. (2009). Westside Barbell Bench Press Manual (pp. 12–15). Westside Barbell. Practitioner source supplied for historical context.
- Castro, A. H., Zangakis, D., & Moir, G. L. (2020). The effects of accentuated eccentric loading on mechanical variables and agonist electromyography during the bench press. Sports, 8, 79. https://doi.org/10.3390/sports8060079 ↩
- Kristiansen, E. L., Larsen, S., & van den Tillaar, R. (2022). The acute effect of accentuated eccentric overloading upon kinematics and myoelectric activity in the eccentric and concentric phase of a traditional bench press. Sports, 10, 6. https://doi.org/10.3390/sports10010006 ↩
- Doan, B. K., Newton, R. U., Marsit, J. L., Triplett-McBride, N. T., Koziris, L. P., Fry, A. C., et al. (2002). Effects of increased eccentric loading on bench press 1RM. Journal of Strength and Conditioning Research, 16, 9–13. https://pubmed.ncbi.nlm.nih.gov/11834100/ ↩
- Ojasto, T., & Häkkinen, K. (2009). Effects of different accentuated eccentric load levels in eccentric–concentric actions on acute neuromuscular, maximal force and power responses. Journal of Strength and Conditioning Research, 23, 996–1004. https://doi.org/10.1519/JSC.0b013e3181a2b28e ↩
- van den Tillaar, R., & Kwan, K. (2020). The effects of augmented eccentric loading upon kinematics and muscle activation in bench press performance. Journal of Functional Morphology and Kinesiology, 5, 8. https://doi.org/10.3390/jfmk5010008 ↩
- Yang, J., Nagatani, T., Comfort, P., Kendall, K. L., & Haff, G. G. (2026). Acute effects of accentuated eccentric loading on mean concentric velocity during resistance training: a systematic review and meta-analysis. Sports Medicine - Open, 12, 74. https://doi.org/10.1186/s40798-026-01031-y ↩
- Michalak, M. M., Suchomel, T. J., Greer, B. K., Long, S. A., & Taber, C. B. (2025). The effects of maximal and supramaximal accentuated eccentric loading on the barbell bench press. Journal of Strength and Conditioning Research, 39(10), 1028–1033. DOI: 10.1519/JSC.0000000000005176. Official journal record ↩
- Goldman, P., Taylor, J., Yamamoto, T., Blatney, A. E., Sahni, T. K., Lechner, R. J., et al. (2025). Eccentrically overloaded bench press training: Augmenting strength gains via a novel bench press pad. Scientific Journal of Sport and Performance, 4(4), 480–490. https://doi.org/10.55860/JCDL3612 ↩