OBJECTIVES: In 1995, a cross sectional survey was conducted in 7 Swiss hospitals to estimate the incidence of percutaneous injuries among nurses, surgeons, anesthetists and domestic personnel, and to describe the circumstances of these injuries and the reporting process within the hospital. METHODS: An anonymous questionnaire was distributed and filled out on-site in the case of nursing staff and domestic personnel, and was sent by post to physicians (anesthetists and surgeons). Participants were asked to report in detail on percutaneous injuries of the last workday and the last working month (nurses and physicians), and of the last month and the last year for domestic personnel. The overall response rate was 72%, representing a total of 3116 health care workers. RESULTS: The annual incidence rates of percutaneous injury with material contaminated with blood or other biological fluids were calculated by type of worker for the two available units of time. For nurses, the incidence was 0.49 and 2.23, for surgeons 4.28 and 11.05, for anesthetists 2.11 and 3.14, and for domestic personnel 0.11 and 0.17 respectively. Most of the injuries occurred in a "normal" situation (no emergency, no stress, no fatigue) and were described as avoidable. Compliance with universal precautions was not optimal and declaration rates within the hospital rather low (nurses 39.7%, physicians 3.4%, domestic personnel 87.9%). CONCLUSION: Percutaneous injuries with blood-contaminated material are frequent in health care workers, and are not always adequately assessed because of under-reporting of accidents within the hospital. This may result in underestimation of current occupational exposure of health care workers to HIV and other blood-borne viruses.