To estimate the risk of experiencing clinical features of cocaine dependence within 1-2 years of starting cocaine use, and to examine whether crack smoking might increase this risk.A national sample of recent-onset cocaine users was identified within public data files of the National Household Surveys on Drug Abuse (NHSDA) for the years 1995 through 1998. The sample included 572 recent-onset users of cocaine HCl powder but not crack, and 190 recent-onset users of crack, some of whom had also started use of cocaine powder no more than 23 months prior to assessment. A separate group of 93 recent-onset crack users was identified; this comparison group had started using cocaine HCl powder 2+ years before assessment. Cocaine dependence was assessed via seven standardized questions about clinical features experienced within 12 months of assessment, such as feeling unable to cut down. Multivariate response regressions were used to evaluate crack-associated excess risk and clinical profiles of cocaine dependence.Among persons who had recently started to use cocaine HCl powder but not crack cocaine, about 5-12% experienced clinical features of cocaine dependence. Most clinical features occurred 2-3 times more often among crack smoking users as compared to those using powder only, even with statistical adjustment for frequency of cocaine use ( P<0.01). This crack-associated excess risk is more prominent for several clinical features of cocaine dependence, including tolerance associated with cocaine use and narrowed behavioral repertoire attributed to cocaine use.This new epidemiological evidence suggests that crack-smoking may increase risk of cocaine dependence once cocaine use starts, but we cannot rule out the possibility that crack users start out with a greater susceptibility to become cocaine dependent.