Sprint start performance has previously been quantified using several different measures. This study aimed to identify whether different measures could influence the performance-based ranking within a group of 12 sprinters, and if so, to identify the most appropriate measure. None of the 10 performance measures ranked all sprinters in the same order; Spearman's rho correlations between different block phase measures ranged from 0.50 to 0.94, and between block phase measures and those obtained beyond block exit from 0.66 to 0.85. Based on the consideration of what each measure quantifies, normalised average horizontal external power was identified as the most appropriate, incorporating both block velocity and the time spent producing this velocity. The accuracy with which these data could be obtained in an externally valid field setting was assessed against force platform criterion data. For an athlete producing 678 +/- 40 W of block power, a carefully set-up manual high-speed video analysis protocol produced systematic and random errors of +5 Wand +/- 24 W, respectively. Since the choice of performance measure could affect the conclusions drawn from a technique analysis, for example the success of an intervention, it is proposed that external power is used to quantify start performance.