Kindergarten "redshirting"--the decision to delay a child's kindergarten entry--may have consequences not only for the redshirted child but for other children whose grade cohort is affect. We use two nationally-representative datasets to estimate the prevalence of redshirting, its relationship to observable demographic and developmental characteristics, and its impact on the composition of kindergarten cohorts both at the national and school level. We find that between 4 and 5.5 percent of children delay kindergarten, a lower number than typically reported. Male, white and high SES children are most likely to delay kindergarten. However, given the low redshirting rates nationwide, redshirting has only modest effects on the magnitude of national kindergarten-entry achievement gaps. We find substantial variation in practices across schools, with schools serving larger proportions of white and high-income children having far higher rates of delayed entry. Implications for policy are discussed. (author abstract)