摘要:The aim of the present study was to compare and correlate training impulse (TRIMP) estimates proposed by Banister (TRIMPBanister), Stagno (TRIMPStagno) and Manzi (TRIMPManzi). The subjects were submitted to an incremental test on cycle ergometer with heart rate and blood lactate concentration measurements. In the second occasion, they performed 30 min. of exercise at the intensity corresponding to maximal lactate steady state, and TRIMPBanister, TRIMPStagno and TRIMPManzi were calculated. The mean values of TRIMPBanister (56.5 ± 8.2 u.a.) and TRIMPStagno (51.2 ± 12.4 u.a.) were not different (P > 0.05) and were highly correlated (r = 0.90). Besides this, they presented a good agreement level, which means low bias and relatively narrow limits of agreement. On the other hand, despite highly correlated (r = 0.93), TRIMPStagno and TRIMPManzi (73.4 ± 17.6 u.a.) were different (P < 0.05), with low agreement level. The TRIMPBanister e TRIMPManzi estimates were not different (P = 0.06) and were highly correlated (r = 0.82), but showed low agreement level. Thus, we concluded that the investigated TRIMP methods are not equivalent. In practical terms, it seems prudent monitor the training process assuming only one of the estimates.