The emergence of drug resistance remains a major problem for the treatment of HIV-infected patients. However, the variety of mutational patterns that evolve in clinical practice have made the application of resistance data to clinical decision-making challenging. Despite (or because of) an abundance of drug-resistance data from disparate sources, there is only limited information available describing the patterns of drug resistance which usually appear in the clinic. Here we attempt to address this issue by reviewing HIV drug resistance in the population of patients failing antiretroviral therapy in British Columbia, Canada from June 1996 to December 2003 as an example. Our findings suggest that, although hundreds of mutations have been associated with resistance, relatively few key mutations occur at a high frequency. For example, only the nucleoside reverse transcriptase inhibitor (NRTI) mutations M184V, M41L T215Y, D67N, K70R and L210W, non-nucleoside reverse transcriptase inhibitor (NNRTI) mutations K103N and Y181C, and protease inhibitor (PI) mutation L90M, occur in more than 10% of samples tested for resistance in this population. The introduction of new drugs allows for the selection of new mutations. Trends in the prevalence of resistance-associated mutations have generally followed trends in drug usage, but have not always mirrored them. The phenomenon of cross-resistance can play an important role in the efficacy of new antiretroviral agents, even before they become available. The extent of this cross-resistance depends in part on the prevalence of specific mutations in the population of individuals who have previously received antiretroviral therapy. Hence there is a need to determine which mutations are prevalent in the treated population. The tremendous capacity of HIV to adapt means that common resistance pathways are likely to change over time, and new pathways to resistance are likely to continue to be discovered in the future.