The association between levels of marijuana use and last year dependence is investigated in a nationally representative sample of adolescents and adults, who used marijuana within the last year (n = 9284). Data are aggregated from three surveys (1991-1993) of the National Household Survey on Drug Abuse. A proxy measure of DSM-IV dependence criteria was developed from self-reported symptoms of dependence and drug-related problems. Both frequency and quantity of marijuana use within the last year are linearly associated with the logit of the probability of being dependent on marijuana. The associations vary significantly by age but not gender. Adolescents are dependent at a lower frequency and quantity of use than adults: the differences diverge as level of use increases. Twice as many adolescents as adults who used marijuana near-daily or daily within the last year were identified as being dependent (35% versus 18%). Frequency and quantity of use each retained a unique effect on dependence, but frequency appeared to be more important than quantity in predicting last year dependence. These results provide insight into the processes underlying the age and sex differentials observed in the prevalence of marijuana dependence. The implications of the findings for the epidemiology of marijuana use and dependence are discussed.