Efavirenz resistance during HIV-1 treatment failure is usually associated with the reverse transcriptase mutation K103N. L100I, V108I, or P225H can emerge after K103N and increase its level of efavirenz resistance. K103N + L100I is the most drug-resistant of the double mutants but is the least common clinically. We hypothesized that differences in replication efficiency, or fitness, influence the relative frequencies of these secondary efavirenz resistance mutations in clinical isolates. We measured fitness of each secondary mutant introduced into HIV(NL4-3), alone and in combination with K103N, using growth competition assays in H9 cells. In the absence of efavirenz, the fitness of V108I was indistinguishable from wild type. K103N, L100I, and P225H were minimally, but consistently, less fit than wild type. K103N + L100I had a greater reduction in fitness and was less fit than K103N + V108I and K103N + P225H. The fitness defect of K103N + L100I relative to K103N was completely compensated for by the addition of the nucleoside resistance mutation L74V. In the presence of efavirenz, L100I was less fit than K103N, and K103N + L100I was more fit than K103N + V108I. Our studies suggest the primary driving force behind the selection of secondary efavirenz resistance mutations is the acquisition of higher levels of drug resistance, but the specific secondary mutations to emerge are those with the least cost in terms of replication efficiency. In addition, nucleoside and NNRTI resistance mutations can interact to affect HIV replication efficiency; these interactions may influence which mutations emerge during treatment failure. These studies have important implications for the design of more durable NNRTI-nucleoside combination regimens.