Objectives a) To evaluate the frequency, types, severity, and morbidity of iatrogenic complications; b) determine associated factors that favor iatrogenic complications; and c) suggest new or more efficient protective measures that may be taken to improve patient safety. Design Prospective, observational study. Setting Two ICUs in France. Patients and Methods The study included 382 patients (age ≥15 yrs; 400 consecutive admissions). Patients were monitored by two physicians in each ICU to assess all iatrogenic complications occurring during their ICU stay, with the exception of adverse effects of drugs. An iatrogenic complication was defined as an adverse event that was independent of the patient's underlying disease. Results We observed 316 iatrogenic complications in 124 (31%) of the 400 admissions. Of these iatrogenic complications, 107 (in 53 [13%] of the 400 admissions) complications, were major, three leading to death. Severe hypotension, respiratory distress, pneumothorax, and cardiac arrest represented 78% of the major iatrogenic complications. Fifty-nine percent of the major iatrogenic complications had clearly identified associated factors. Human errors accounted for 67% of these factors. Patients >65 yrs (adjusted odds ratio = 2.6,95% confidence interval: 1.4 to 4.9) and those patients admitted with two or more organ failures (adjusted odds ratio = 4.8, 95% confidence interval: 2.5 to 9.2) were more likely to develop major iatrogenic complications. High or excessive nursing workload also led to an creased risk of major iatrogenic complications Persistent morbidity, secondary to iat complications at the time of discharge, present in five survivors. The risk of ICU dead was about two-fold higher for the patients with major iatrogenic complications than in the re maining patients after adjusting for the Or System Failure Score and the prognosis of the disease (relative risk = 1.92, 95% confidence terval: 1.28 to 2.56). Conclusions Major iatrogenic complications were frequent, associated with increased bidity and mortality rates, related to high excessive nursing workload, and were often ondary to human errors. To improve pati safety in our ICUs, preventive measures sho be targeted primarily on the elderly and the most severely ill patients. Special attent should be given to improving the organization workload and training, and promoting w use of noninvasive monitoring. (Crit Care Med 1993; 21:40–51)