Hepatitis C virus (HCV) and hepatitis B virus (HBV) co-infection may differentially influence HIV treatment duration and outcome. This was assessed at The Ottawa Hospital Immunodeficiency Clinic in first-time highly active antiretroviral therapy (HAART) recipients visited between January 2000 and December 2004. Of 968 patients, 526/700 (75%) HIV, 173/230 (75%) HIV-HCV and 30/38 (79%) HIV-HBV-infected patients initiated HAART. Co-infected patients stopped treatment sooner (HBV - 10 months, HCV - 9 months) than HIV mono-infected (17 months) (P<0.001). Injection drug history predicted shorter treatment duration (odds ratio [OR]1.59, P<0.001). Use of non-nucleoside-reverse-transcriptase-inhibitor-containing HAART (OR 0.76, P<0.01) and low-dose ritonavir (<400 mg twice daily)-based HAART (OR 0.83, P = 0.06) predicted longer treatment duration. HCV co-infection did not predict duration of therapy (OR 1.19, P=0.19) once controlled for by these three variables. Poor adherence was a major explanation for eventual treatment interruption in those with HIV-HCV (22% versus 5% in HIV alone; P<0.001) as was substance abuse (7% versus < 1% in HIV; P<0.001). Metabolic complications resulted in HAART interruption in HIV mono-infection (8%) but not with HBV or HCV co-infection (both <1%; P<0.001). Antiretroviral selection is critical to the longevity of initially prescribed regimens, irrespective of viral hepatitis co-infection. Attention to this and strategies targeting substance abuse and adherence in HIV-HCV are predicted to increase the duration of HAART.