Skip to main content
. 2019 Apr 11;49(7):1007–1030. doi: 10.1007/s40279-019-01101-0

Table 2.

Summary of studies exploring the effects of caffeine between trained and untrained individuals

Reference Sample Caffeine dose Performance metric Main findings
Astorino et al. [97] 8 endurance-trained and 8 ‘active’ young men 5 mg/kg 10-km cycling time trial Caffeine ingestion reduced the time necessary to complete 10-km of cycling in endurance-trained but not in ‘active’ men
Boyett et al.a [83] 7 endurance-trained and 7 untrained young men 6 mg/kg Isokinetic knee extension and 3-km cycling time trial For cycling time trial, the differences in responses to caffeine and placebo ingestion in the morning training sessions were ‘unclear’ between the groups; for the two evening conditions, following caffeine ingestion, untrained individuals ‘likely’ experienced greater reductions in time necessary to complete 3-km of cycling than trained individuals; for isokinetic peak torque, the comparisons were either ‘trivial’ or ‘unclear’
Brooks et al. [109] 7 resistance-trained and 7 untrained young men 5 mg/kg Weight lifted and force produced in the 1 RM Smith machine squat Caffeine ingestion improved 1 RM weight lifted in untrained but not in resistance-trained men; no between-group differences were observed for force production
Collomp et al. [24] 7 trained swimmers and 7 untrained swimmers (young men and women) 250 mg 1600-m swimming for the trained swimmers and 400-m for the untrained Caffeine ingestion improved swimming velocity in trained but not in untrained participants
O’Rourke et al. [108] 15 young well-trained and 15 recreational runners (sex was not specified) 5 mg/kg 5-km running time trial Caffeine ingestion reduced time necessary to complete 5-km of running in both well-trained and recreational runners
Porterfield et al. [107] 10 endurance-trained and 10 untrained young men 5 mg/kg Cycling time to exhaustion Caffeine ingestion did not improve time to exhaustion either in endurance-trained or untrained men

1 RM 1 repetition maximum

aData were analyzed using the magnitude based inferences approach