To investigate the effect of the ingestion of arginine aspartate (AA) in the decrease of the exercise-induced accumulation of ammonia in plasma, 11 voluntary subjects took part in a cross-over study where AA effect was tested against placebo. Both treatments were randomly administered in a double-blind procedure. To ensure the subjects would be able to present reproducible exercise-testing results during repetitive sessions, they were involved before the experiment in a cycle ergometer training program during 8 weeks. This training determined a significant 14% increase (P less than 0.001) in maximal oxygen uptake (VO2 max). The treatments were administered during 10 days and the two treatments were separated by a 10 day-wash-out period. A 45 min-cycle ergometer test was performed at 80% VO2 max during the 10th day of each treatment to measure plasma ammonia (p[NH4+]) and total blood lactate (b[lact]) concentrations at rest and at the 15th, 30th and 45th min of exercise (determinations of changes from rest; delta p[NH4+] and delta b[lact]). Both concentrations were unchanged between AA and placebo at rest but a significant lesser delta p[NH4+] was found under AA at the 15th min of exercise only (P less than 0.05). On the other hand, an order effect was found for delta p[NH4+] between the two periods of randomized treatment that was interpreted as a remaining training effect. This effect was highly significant at the 30th and 45th min of exercise (P less than 0.001). It was concluded that AA effect was minor with regard to the training effect. As it was not located at the same time of exercise, AA effect would not consequently have the same functional origin (postulated increase in the peripheral clearance of ammonia) than those of training (decrease in muscle production of ammonia).