Velocity Loss, Fatigue & Bench-Press Adaptation

What velocity loss measures

The usual calculation is: 100 × (reference velocity − current velocity) ÷ reference velocity. If the fastest valid rep is 0.50 m/s and a later rep is 0.375 m/s, the loss is 25%. Some systems reference the first repetition; others use the fastest rep reached after the athlete settles into the set. The choice must be consistent because an unusually slow first rep or an unusually fast rebound changes the stop point.

Sánchez-Medina and González-Badillo showed that within-set velocity loss is strongly related to acute neuromuscular fatigue across resistance exercises [1]. Later bench and squat work found strong relationships between relative velocity loss and the percentage of repetitions completed, but the number of repetitions possible at a given load varied considerably among individuals [2][3]. Velocity loss improves effort standardization; it does not make athletes identical.

The bench-specific dose–response evidence

The four-threshold 2020 trial

Pareja-Blanco and colleagues randomized 64 resistance-trained young men to 0%, 15%, 25% or 50% velocity-loss thresholds in an eight-week bench program. The investigators assessed dynamic and isometric performance, a progressive-loading test, fatigue resistance, pectoralis major cross-sectional area and surface EMG. Load progression and velocity feedback standardized the sessions; allowed loss determined how many repetitions each group completed [4].

All four groups improved dynamic strength, with no significant group-by-time interaction for those strength measures despite large volume differences. Pectoralis cross-sectional area did differ by group: the 50% condition increased more than 0%. Only the 15% group significantly increased peak pectoralis RMS EMG, the 0% group showed an early rate-of-force-development change, and 25% and 50% favored later RFD measures. This was not a linear “more fatigue, more strength” result. High loss maximized muscle size in this protocol, while lower loss produced strength efficiently and distinct neuromuscular changes.

Two intensity-specific extensions

Rodiles-Guerrero and colleagues assigned 50 trained men to the same four thresholds for eight weeks at 55–70% 1RM. All groups except 0% significantly improved 1RM; the 25% condition produced the largest 1RM and load–velocity changes. Only 25% significantly increased pectoralis EMG, while only 50% significantly increased cross-sectional area. The authors favored about 25% for broad adaptation at those moderate loads [5].

With lighter 40–55% loads, a later trial randomized 46 trained men and held frequency, three sets and four-minute rests constant. There was no significant group-by-time strength interaction, though effect sizes again formed an inverted U with 25% generally largest. The total repetition cost was stark: 48 at 0%, 357 at 15%, 547 at 25% and 876 at 50%. The 50% group gained no extra strength despite doing more than 18 times the repetitions of 0% [6]. “Efficient” depends on the outcome; for 1RM, the high-loss work was not rewarded.

The 2026 intensity-by-threshold trial

Pareja-Blanco and colleagues randomized 158 resistance-trained men into 12 groups: three intensity bands (40–55%, 55–70% or 70–85% 1RM) crossed with four thresholds (0%, 15%, 25% or 50%). Training lasted eight weeks, twice weekly, with three sets, four-minute rests, a Smith-machine bench, approximately one second on the chest and maximal intended concentric velocity. Individual load–velocity relationships set training loads, and immediate feedback reinforced intent [7].

There was no three-way intensity-by-loss-by-time interaction, so no single cell can be crowned an absolute winner. Across thresholds, 70–85% produced the greatest strength and pectoralis cross-sectional-area gains. Across intensities, the raw mean 1RM changes were 5.9 kg at 0%, 8.2 kg at 15%, 11.2 kg at 25% and 9.2 kg at 50%; the velocity-loss-by-time interaction was significant, and 25% outperformed 0%. Cross-sectional area rose most at 50%, which exceeded 0% and 15%, and 50% produced the greatest maximal number of repetitions.

The volume gradient again mattered. Average training repetitions were about 48 at 0%, 220 at 15%, 336 at 25% and 534 at 50%. A 50% threshold was effective for size and endurance, but 25% produced the largest 1RM gain with roughly 198 fewer repetitions. Statistical significance establishes a group difference under this protocol; the raw kilogram changes communicate its practical magnitude. Baseline 1RM differed slightly across groups, so the bars below are descriptive mean changes, not adjusted causal effects.

Read threshold and intensity together

This is a decision map synthesized from bench-specific trials, not a universal prescription or direct reproduction of one study. Threshold effects depend on intensity, device, exercise, athlete and program duration.

Statistical signal versus coaching meaning

A non-significant group interaction does not prove protocols are identical; it means the study did not resolve a difference beyond its uncertainty. Conversely, a significant interaction does not make 25% optimal for every athlete or goal. The 2026 trial’s 5.3 kg raw gap between 25% and 0% is practically visible, but it was achieved with about seven times the repetitions. A time-limited in-season athlete may prefer the smaller gain per block with lower fatigue; a hypertrophy block may accept the higher volume.

Technique can also end the set before the numeric threshold. Duffey and Challis documented systematic bench kinematic changes as fatigue accumulated, including altered bar path and joint coordination [8]. If touch point, pause, range, contact points or bar path leave the prescribed corridor, the repetition is no longer the task used to establish the reference velocity. A velocity device should never overrule a clearly invalid rep.

Failure is not required for strength or hypertrophy in every context. A meta-analysis comparing failure and non-failure training found no general advantage for momentary failure when volume was not otherwise controlled, with substantial protocol heterogeneity [9]. Velocity loss is useful precisely because it offers several stopping points before failure rather than reducing every set to “easy” or “all out.”

Efficiency has to be named, not assumed

A low-loss prescription is clearly efficient if the denominator is repetitions or minutes spent accumulating fatigue. It may be less efficient if an athlete needs more sets, more weekly exposures or a longer block to reach a hypertrophy target. Conversely, 50% loss generated the largest pectoralis cross-sectional-area and repetition-endurance responses in the largest trial, but it also required roughly 534 training repetitions versus 48 at 0%. Calling either strategy “best” without naming the outcome hides the tradeoff exposed by the data [7].

Strength efficiency should also be separated from maximum observed strength gain. In the 2026 study, 25% produced the largest raw mean 1RM change and used fewer repetitions than 50%, but 0% still improved strength with a fraction of the work. The practical ratio changes again when session setup, athlete monitoring and equipment time are considered. A team coach may accept slightly less average gain to preserve time and readiness across a roster; an off-season lifter may accept moderate fatigue for the larger group-average response. Neither decision contradicts the trial.

Finally, the programmed percentage is not the received dose unless the reference repetition is valid. If the first rep is slow because of a poor handoff, a 25% threshold arrives prematurely. If a bounce produces an artificial fastest rep, it arrives too soon for a different reason. Record the fastest valid rep, repetitions completed, terminal velocity, reason for stopping and next-session performance. Those fields distinguish a threshold that managed fatigue from one that merely reacted to measurement noise.

Three 2026 extensions of the threshold evidence

Trained women: outcomes separated by threshold

In 49 trained women completing eight weeks at 70%–85% 1RM, VL50 improved 1RM more than VL0 and was the only condition to increase measured triceps thickness; VL25, not VL50, produced some favorable force–time and EMG changes [10]. The trial is valuable because it broadens a literature dominated by men and shows why “best threshold” depends on the target outcome.

Prior fatigue weakens the repetition–loss relationship

In 14 men, Rodiles-Guerrero and colleagues tested a set to failure at 60% 1RM after prior sets stopped at 0%, 20%, 40% or 60% velocity loss. Velocity loss remained useful, but its relationship with repetitions remaining degraded as prior fatigue increased; reliability was better under low-to-moderate fatigue [11]. A threshold cannot erase what happened earlier in the session.

Variable resistance can preserve the signal

Ocaña-García and colleagues reported strong relationships between repetitions completed and velocity loss during band-resisted bench pressing in 26 participants (reported R² .910–.935) [12]. That supports measurement feasibility under the tested band setup, not automatic interchangeability with straight-weight thresholds.

Measurement quality before threshold precision

Linear position transducers generally provide the most direct bar-displacement measurement, while camera systems, inertial units and cable devices estimate velocity differently. Device validity can change by exercise and speed. Use the same device, attachment point, metric and filtering method across a block. Do not mix mean and peak velocity or compare values from different brands as if they share an algorithm.

• Warm up until technique and velocity are stable, not fatigued.
• Standardize grip, touch, pause, range, arch and command.
• Use maximal intended concentric velocity on every counted rep.
• Set a minimum absolute velocity or technical stop alongside percentage loss.
• Review outliers before changing the program; one loose cable or misgroove is not adaptation.

Programming by outcome

These are starting bands inferred from group evidence, not clinical or universal prescriptions. Choose relative intensity first. For a strength block, the 2026 data support placing much of the work at 70–85% 1RM and using a moderate loss threshold rather than trying to make light sets exhaustive. For hypertrophy, a higher threshold can be rational, but weekly set count, recovery and symptoms still constrain the dose.

Use a two-part stop rule: stop at the planned velocity loss or at the first invalid repetition, whichever comes first. If athletes repeatedly stop for technique before velocity, lower the load or improve the standard. If day-to-day first-rep velocity is unusually low, consider readiness and reduce load rather than forcing the planned number of repetitions.

What the AMM Studies Show About Bar Velocity

In the acute AMM crossover, 10 resistance-trained men performed five repetitions at 70% 1RM on a standard bench and the Launch Pad. Mean concentric velocity was 0.40 ± 0.09 versus 0.47 ± 0.09 m/s, and peak velocity was 0.57 ± 0.11 versus 0.66 ± 0.09 m/s; both differences were significant [13]. The experiment measured bar velocity across the tested condition. It did not calculate a threshold-based repetition-by-repetition velocity-loss metric.

The four-week trial used five sets to failure, three times weekly, with 150% eccentric resistance in both groups; the Launch Pad group gained more 1RM than the conventional-bench group [14]. The eight-week collegiate football trial used matched periodized programs and reported larger 1RM, NFL-225 and seated medicine-ball-throw improvements with the Launch Pad [15]. Those trials evaluated training outcomes rather than velocity loss as a mediator.

Together, the studies show higher measured bar velocity in the acute condition and larger strength or performance gains in two training comparisons. Joint Ops remains outside this published evidence chain. The scientific relevance is a clear, testable bridge from higher measured bar velocity to future repetition-by-repetition and mediator-focused studies.