The universal precautions recommended by the US Centers for Disease Control (CDC), Atlanta, for the prevention of HIV (human immunodeficiency virus) transmission to health care workers are widely accepted, despite little documentation of their effectiveness and efficiency. We reviewed the evidence on the risk of HIV transmission to hospital workers and the effectiveness of the universal precautions. We also evaluated the costs of implementing the recommendations in a 450-bed acute care teaching hospital in Hamilton, Ont. On the basis of aggregated results from six prospective studies the risk of HIV seroconversion among hospital workers after a needlestick injury involving a patient known to have AIDS (acquired immune deficiency syndrome) is 0.36% (upper 95% confidence limit 0.67%); the risk after skin and mucous membrane exposure to blood or other body fluids of AIDS patients is 0% (upper 95% confidence limit 0.38%). We estimated that 0.038 cases of HIV seroconversion would be prevented annually in the study hospital if the CDC recommendations were followed. The incremental cost of implementing the universal precautions was estimated to be about $315,000 per year, or over $8 million per case of HIV seroconversion prevented. If all HIV-infected workers were assumed to have AIDS within 10 years of infection the of the program would be about $565,000 per life-year saved. When less conservative, more probable assumptions were applied the best estimate of the implementation cost was $128,862,000 per case of HIV seroconversion prevented. The universal precautions implemented in the study hospital were not found to be efficacious or cost-effective. To minimize the already small risk of HIV transmission in hospitals the sources of risk of percutaneous injury should be better defined and the design of percutaneous lines, needles and surgical equipment as well as techniques improved. Preventive measures recommended on the basis of demonstrated efficacy and aimed at routes of exposure that represent true risk are needed.