An independent quality assurance program has been established by the Division of AIDS, National Institute of Allergy and Infectious Diseases, for monitoring virologic assays performed by nearly 40 laboratories participating in multicenter clinical trials in the United States. Since virologic endpoints are important in evaluating the timing and efficacy of therapeutic interventions, it is imperative that virologic measurements be accurate and uniform. When the quality assurance program was initially created, fewer than 40% of the laboratories could consistently recover human immunodeficiency virus (HIV) from peripheral blood mononuclear cells (PBMCs) of HIV-infected patients. By comparing coculture procedures in the more competent laboratories with those in laboratories who were struggling to isolate virus, optimal conditions were established and nonessential reagents and practices were eliminated. Changes were rapidly introduced into a laboratory when experience dictated that such modifications would result in a favorable outcome. Isolation of HIV was enhanced by optimizing the numbers and ratios of patient and donor cells used in cultures, by standardizing PBMC separation procedures, by using fresh rather than frozen donor PBMCs, by processing whole blood within 24 h, and by using natural delectinated interleukin 2 instead of recombinant interleukin 2 products in existence at that time. Delays of more than 8 h in the addition of phytohemagglutinin-stimulated donor cells to freshly separated patient PBMCs reduced recovery. Phytohemagglutinin in cocultures and the addition of Polybrene and anti-human alpha interferon to media were not important in HIV isolation. The introduction of a consensus protocol based on this information brought most laboratories quickly into compliance. In addition, monthly monitoring has successfully maintained proficiency among the laboratories, a process that is critical for the scientific integrity of collaborative multicenter trials. Problems which might not be appreciated for months are now being resolved early, before data can be compromised unknowingly. This consensus protocol is recommended for any laboratory attempting to isolate HIV for the purpose of standardizing recovery and for accessing virologic endpoints in clinical trials.