Our study sought to determine the incidence of weekly marijuana use among HIV-infected and uninfected women, to identify correlates of weekly marijuana use, and to test its association with stage of HIV disease and type of HIV treatment received. A total of 2059 HIV-positive and 569 HIV-negative women from 6 sites were recruited between 1994 and 1995 and followed through 2000. After excluding women who reported weekly marijuana use at baseline, 2050 women were included in the analysis. The incidence rate for initiating marijuana was calculated and survival analysis was performed to determine the correlates of initiating weekly marijuana use. Three hundred and three women initiated weekly marijuana use within 5.5 years of the baseline visit, yielding a cumulative incidence (CI) of 14.8%. There was no significant difference in weekly marijuana use initiation between HIV-infected (CI = 14.5%) and HIV-uninfected women (CI = 16.0%). Younger age and having more sex partners was associated with incident weekly marijuana use among both infected and uninfected women. While undetectable viral load was associated with lower incidence rate (p < 0.001, RH = 0.44) and wasting syndrome with higher incidence (p < 0.01, relative hazard [RH] = 3.1), CD4 count was not. Compared to receiving no AIDS treatment at all, women who received basic combination antiretroviral therapy had significantly higher incidence of weekly marijuana use (p < 0.001, RH = 1.93), while highly active antiretroviral therapy (HAART) receivers had significantly lower incidence (p < 0.001, RH = 0.24). In summary, among HIV-infected women, the incidence of weekly marijuana use was associated with only one marker of HIV disease stage and HAART was associated with lower initiation rate of weekly marijuana use.