The purpose of this study was (1) to examine the value of a verification phase (VER) in a peak testing protocol and (2) to assess the reliability of peak physiological variables in wheelchair athletes. On two separate days, eight tetraplegic (TETRA), eight paraplegic (PARA) and eight non-spinal cord-injured (NON-SCI) athletes performed treadmill ergometry, consisting of a graded exercise test to exhaustion (GXT) followed by a VER. Peak oxygen uptake (.VO2peak) was compared (1) between GXT and VER and (2) between test days. .VO2peak did not differ between GXT and VER (P = 0.27), and coefficients of variation between GXT and VER were in the range of 2.9 and 6.4 % for all subgroups. Coefficients of variation of .VO2peak between test days were 9.3 % (TETRA), 4.5 % (PARA) and 3.3 % (NON-SCI). It is therefore concluded that whilst a VER can be used for a more robust determination of .VO2peak, a deviation of up to ~6 % between GXT and VER should be considered as acceptable. For between-day analyses, relatively large changes in .VO2peak are required to confirm "true" differences, especially in TETRA athletes. This may be due to their lower aerobic capacity, which results in a larger relative variation compared with the other subgroups.