REF-1647 · Capelli et al., 2016 · European Journal of Applied Physiology, Vol. · Paper · PubMed

Modelled from the best masters track times (200-3000 m, men aged 30-80), maximal aerobic power began to fall at 45 years and then fell by about 16% per decade, and anaerobic capacity by about 11% per decade from the second half of the fifth decade

Estimates from an energetics model, not measurements in riders

Main caveat: MAP and anaerobic capacity are model estimates from race times, not laboratory measurements; the model's assumptions and validation are not in the abstract.
weak evidence: opinion or lab dataMeasured on trained to highly trained athletesRead from the abstract only

Capelli and colleagues used the best times of male masters cyclists in 200-3000 m track events to estimate how maximal aerobic power (MAP) and anaerobic capacity (AnS) decay with age.

For each age decade between 30 and 80 years, an iterative procedure found the MAP and AnS values that minimised the gap between the metabolic power needed to cover a distance in the recorded time and the maximal metabolic power that could be held steady through the race.

Estimated MAP started to fall at 45 years and then fell on average by about 16% per decade, a rate the authors say matches other groups of masters athletes; AnS seemed to fall by about 11% every 10 years from the second half of the fifth decade.

The authors tentatively attribute the MAP loss, which occurred despite the athletes' active lives, to a gradual impairment of maximal oxygen delivery and/or its use in muscle, and the AnS loss to the loss of muscle mass after the fifth decade, poorer anaerobic energy pathways or a lower capacity to recruit motor units in maximal efforts.

They suggest the approach suits other forms of locomotion where the relation between performance time and metabolic power is known.

Participants

Male masters track cyclists, represented by their best performance times in 200-3000 m events, grouped by decade from 30 to 80 years.
Not in the abstract: the number of athletes and performances, the competitions or record lists used, and the years covered.

Protocol

Data: best masters performance times over 200-3000 m on the track, by age decade.
Model: for each decade, MAP and AnS chosen by iteration to minimise the difference between (1) the metabolic power required to cover distance d in time t and (2) the maximal metabolic power maintained at a constant level through the event.
Not in the abstract: the energy-cost assumptions, how the aerobic and anaerobic contributions were modelled over time, and any validation against measured MAP or anaerobic capacity.

Key Results

Maximal aerobic power: started to decrease at 45 years; thereafter an average fall of about 16% per decade.
Anaerobic capacity: fell by about 11% every 10 years from the second half of the fifth decade.
Absolute MAP or AnS values per decade are not in the abstract.
A published comment (Borrani & Millet, PMID 27637590) questioned the age at which the decline starts; its content is not in its PubMed record.

Practical Application

Expect the ceiling to fall with age even at the top: in this model, aerobic power held until about age 45 and then dropped about 16% per decade, and anaerobic capacity about 11% per decade from the second half of the fifth decade of life. Use it as a rough expectation for the best male masters riders, not as a forecast for one person.
Inference, not tested here: a masters rider comparing today's power with a decade ago should expect some decline even with consistent training; in the base's longitudinal runner data, continued hard training slowed the loss of VO2max but did not stop it (REF-595).
The authors' explanations (oxygen delivery and use, muscle mass, motor-unit recruitment) are hypotheses here; the model estimated capacities and tested no mechanism.
Not settled by this design / not from this paper: whether training changes the slope, women, long road or endurance events, individual trajectories rather than age-group bests, and the exact onset age, which a published comment disputed.

Limitations

MAP and anaerobic capacity are model estimates from race times, not laboratory measurements; the model's assumptions and validation are not in the abstract.
Best times per age decade compare different athletes at each age; they describe the top of each age group, not how one person ages.
Men only; track events of 200-3000 m only.
The number of athletes and performances is not in the abstract.
An erratum exists (its content is not in its PubMed record), and a published comment disputed the onset age.
Written from the abstract and the PubMed record; full text not read.
How this entry was checked and changed
Why this level

modelling: the inputs were measured (best competition times), but MAP and AnS were estimated by the model, not measured in riders, and the abstract reports no check against measured values; the number of performances and athletes is not in the abstract

Content correction 2026-11-07
2026-11-07: added from the abstract and the PubMed record; level C (modelling of performance records; capacities estimated, not measured, and no validation in the abstract).