Abstract
Research Quarterly articles dealing with whether to use the best score or average score as the criterion score for multiple-trial physical performance tests are briefly reviewed. The main purpose of this study is to compare by accepted statistical criteria the best and average scores as estimators of a subject's maximum performance. The results suggest that unless the variance of the measurement error is much larger than the within-subject variance, a rare occurrence, the best score will usually be the better estimator of maximum performance. A consistent estimator of maximum performance is also constructed.