Although evidence is accumulating that needle exchange programs can lower the risk of parenterally transmitted infections, their effectiveness is compromised if they suffer from low client participation. A legal needle exchange in New Haven, CT, has been studied since its inception in November 1990, employing a tracking system to analyze the characteristics of clients participating during the first 20 months of the program. Thirty-four percent of injection drug users who enrolled in the program during the study period made only a single visit. Younger clients were more likely to be in the single visit group. For clients who visited the program more than once, the retention fraction, defined as the ratio of total client-specific observed person-days to full enrollment person-days during the 20-month study period was 67.7 percent, with a median duration of participation of 333 days. Further analysis of the client characteristics, based on surveys completed upon enrollment in the program, revealed several predictors of continuing participation. Most significant were the observations that (a) those injecting for 10 years or longer participated longer than clients who injected drugs for less than 10 years and (b) nonwhite injectors participated longer than whites. The longest duration of participation (median = 501 days) was among nonwhite injectors with 10 years or more of injecting history. There have been limited data on client participation in needle exchange programs. This gap in information must be overcome to allow thorough evaluations of such programs.