We recently reported that HIV-1 resistant to 3-azido-3-deoxythymidine (AZT) is not cross-resistant to 3-azido-2,3-dideoxypurines. This finding suggested that the nucleoside base is a major determinant of HIV-1 resistance to nucleoside analogs. To further explore this hypothesis, we conducted in vitro selection experiments by serial passage of HIV-1 LAI in MT-2 cells in increasing concentrations of 3-azido-2,3-dideoxyguanosine (3-azido-ddG), 3-azido-2,3-dideoxycytidine (3-azido-ddC), or 3-azido-2,3-dideoxyadenosine (3-azidoddA). 3-Azido-ddG selected for virus that was 5.3-fold resistant to 3-azido-ddG compared to wild-type HIV-1 LAI passaged in the absence of drug. Population sequencing of the entire reverse transcriptase (RT) gene identified L74V, F77L, and L214F mutations in the polymerase domain and K476N and V518I mutations in the RNase H domain. However, when introduced into HIV-1 by site-directed mutagenesis, these 5 mutations only conferred ϳ2.0-fold resistance. Single-genome sequencing analyses of the selected virus revealed a complex population of mutants that all contained L74V and L214F linked to other mutations, including ones not identified during population sequencing. Recombinant HIV-1 clones containing RT derived from single sequences exhibited 3.2-to 4.0-fold 3-azido-ddG resistance. In contrast to 3-azido-ddG, 3-azido-ddC selected for the V75I mutation in HIV-1 RT that conferred 5.9-fold resistance, compared to the wild-type virus. Interestingly, we were unable to select HIV-1 that was resistant to 3-azido-ddA, even at concentrations of 3-azido-ddA that yielded high intracellular levels of 3-azido-ddA-5-triphosphate. Taken together, these findings show that the nucleoside base is a major determinant of HIV-1 resistance mechanisms that can be exploited in the design of novel nucleoside RT inhibitors.