Some parents have chosen to delay their children's enrollment in kindergarten by a year because of individual differences in the pace and pattern of children's development. In other situations, some schools recommend delaying a child's school entry based on "readiness" testing or require that some kindergartners attend kindergarten for a second year. Data from the 1993 and 1995 National Household Education Surveys show that about one child in seven either entered kindergarten late or was required to repeat kindergarten. The surveys found similarities between these two categories of children, but also some notable differences pertaining to gender, race, and developmental delays. The surveys found striking differences in later school performance (in grades one and two) between children who were held out of kindergarten and children who repeated kindergarten. The performance of those who had been held out of kindergarten was found to be better in first and second grade than that of children who entered kindergarten at the prescribed age. In contrast, those who were required to repeat kindergarten were doing worse than their first- and second-grade peers. First- and second-graders in 1993 who had repeated kindergarten were more likely than children who had not repeated kindergarten to receive negative feedback from their teachers. When demographic, socioeconomic, and developmental factors were taken into account, the differences in school performance were reduced, but remained significant in the 1993 survey. In the 1995 survey, however, controlling for these background factors essentially eliminated the differences between student who were held out and other first- and second-graders. The same was true of performance differences between the students who had been retained and other students. The surveys did not find evidence that children who may have been at increased risk of experiencing difficulties in school benefited from (or were harmed by) delayed kindergarten entry. The same was true of kindergarten retention. (Interview items from both years, analyses, and logistic regression model coefficients are appended. Contains 44 references.) (HTH)