The aim of the study was to examine whether 1) the power output attained in the last 30 s of a 3-min all-out test (P(end)) correlates with the power output at maximal lactate steady state (P(MLSS)) and whether 2) peak oxygen uptake (VO(2peak)) can be obtained from a 3-min all-out test in well-trained cyclists. 18 cyclists (23 +/- 3 years; 186.1 +/- 6.9 cm; 79.1 +/- 8.2kg; VO(2peak) : 63.2 +/- 5.2 mL . kg(-1) . min(-1)) performed a ramp test, a 3-min all-out test and several submaximal constant 30 min-workload tests at + 15, 0, -15, -30, -45, -60, -75, -90 W of P(end) to obtain P(MLSS). P(MLSS) was significantly lower compared to P(END) (p < 0.001; mean difference: 54 +/- 18 W) with a high correlation (r = 0.93; R(2) = 0.87; p < 0.001) but great intra-individual variability (15-90 W). There were no mean differences between the ramp-VO(2)peak and 3-min all-out cycling VO(2peak) (p = 0.29; mean difference: 133 +/- 514 mL . min(-1)) showing significant correlation (r = 0.60; R(2) = 0.37; p = 0.006) but great intra-individual variability (1057-1312 mL . min(-1)). We therefore suggest that in well-trained cyclists a 3-min all-out test is 1) not sufficient to obtain P(MLSS) and 2) should not be applied to assess VO(2peak).