The study was facilitated by the fact that Florida maintains data on recidivism for all youths released from both nonresidential and residential juvenile programs in the State. In addition, there are a variety of programs, including State-operated and contracted private providers and various security levels. There is broad variation in lengths of stay, ranging from under 3 months to over 18 months. Data were obtained from the Florida Department of Juvenile Justice Information System (JJIS), the Florida Department of Law Enforcement's Florida Criminal Information Center, and the Florida Department of Corrections. The JJIS system was used to identify 16,779 youths released from commitment programs into the community during the 2-year period between July 1, 1998, and June 30, 2000. Length of stay was measured as the number of months spent in the program. This measure was analyzed both at the ratio level and at the ordinal level. "Recidivism" was defined as a subsequent juvenile adjudication, adjudication withheld, or adult conviction for an offense that occurred within 1 year of a juvenile's release into the community or a conditional release program. Separate analyses were conducted on the effects of length of stay on recidivism for juveniles released from residential (institutional) and nonresidential (noninstitutional) programs. A total of 55 nonresidential programs and 185 residential programs were included in the analyses. The analysis showed no consistent relationship between length of confinement and recidivism. Length of stay was significant at the bivariate level for nonresidential and high-risk program releasees; however, in multivariate analyses, the effects of length of stay were only significant for juveniles released from high-risk facilities. The impact of months served for high-risk offenders varied. The shortest lengths of stay within this security level resulted in a decreased likelihood of recidivism. Intermediate periods of confinement, compared to the longest lengths of stay for high-risk juveniles (13 months or more), increased the odds that a juvenile would be readjudicated/convicted. The study hypothesizes a number of competing explanations for these findings and offers suggestions for future research on this subject. 13 references, 7 tables, and 2 figures