OBJECTIVE:: The incidence of certain non-AIDS-defining cancers (NADC) in HIV patients has been reported to have increasedin the combination antiretroviral therapy (ART) era. Studies are needed to directly evaluate the effect of ART use on cancer risk. DESIGN:: We followed 12,872 HIV+ Kaiser Permanente members whose complete ART history was known for incident cancers between 1996-2008. METHODS:: Cancers, identified from SEER-based cancer registries, were grouped as AIDS-defining cancers (ADC), infection-related NADC or infection-unrelated NADC. We also evaluated the most common individual cancer types. Rate ratios (RR) for ART use (yes/no) and cumulativeduration of any ART, PI and NNRTI therapy were obtained from Poisson models adjusting for demographics, pre-treatment or recent CD4 count and HIV RNA levels, years known HIV-infected, prior antiretroviral use, HIV risk, smoking, alcohol/drug abuse, overweight/obesity, and calendar year. RESULTS:: The cohort experienced 32,368person-yrs (py) of ART, 21,249pyof PI therapy, and 15,643pyof NNRTI therapy. The mean follow-up duration was 4.5 years. ADC rates decrease with increased duration of ARTuse [RR/year = 0.61, 95% CI (0.56-0.66)]; the effect was similar by therapy class. ART, PI or NNRTI therapy duration was not associated with infection-related or infection-unrelated NADC [RR/yearART = 1.00 (0.91-1.11) and 0.96 (0.90-1.01), respectively], excepta higher anal cancer risk with longer PI therapy [RR/year = 1.16 (1.02-1.31)]. CONCLUSIONS:: No therapy class-specific effect was found for ADC. ART exposure was generally not associated with NADCrisk, except for long term use of PI, which might be associated with increased anal cancer risk.