Objectives When individuals of different ages are combined into a single group and an ability that varies with age is measured, younger individuals are disadvantaged. This phenomenon is known as a relative age effect (RAE) and has been shown to be widespread in sport and education. Methods In this article, we examine RAEs in a large group of children tested repeatedly on the 20‐m shuttle run, a common test of cardiorespiratory fitness. Following up on an earlier study that measured change with age, we add a growth curve model for change in variance, which makes it possible to derive RAEs for individuals of different ages and ability levels. Results Results show that a 1‐year difference in age is associated with a change in performance of about 0.2 standard deviations. For 1‐year age groups, this gives rise to ranking errors of 4 percentile ranks or less. We also show, however, that these relatively small ranking errors are capable of producing large age differences within groups identified as exceptional. Depending on the level of ability required for selection, children born in the first quarter of the year can be expected to outnumber those born in the last by 1.5 times, 2 times, or more. This finding is consistent with previously reported variation in RAEs at different performance levels. Conclusions Results imply that RAEs are likely to be of relatively minor concern when people are graded or ranked but can produce substantial inequities and misclassifications when people with extremely high or extremely low ability levels undergo selection. Am. J. Hum. Biol., 2014. © 2014 Wiley Periodicals, Inc.