Epidemiology of acute renal failure (ARF): A prospective, multicenter, community-based study. There are very limited data on overall epidemiology of ARF. It is crucial to know the incidence, etiology and clinical features of ARF to promote prevention strategies and to implement adequate resources for the management of this entity. During a nine month period, a collaborative prospective protocol with 98 variables was developed to assess all ARF episodes encountered in the 13 tertiary-care hospitals in Madrid, Spain (covering 4.2 million people of over 14 years of age). ARF was considered when a sudden rise in serum creatinine concentration (S_Cr) to more than 177 µmol/liter was found in patients with normal renal function, or when the sudden rise (50% or more) was observed in patients with previous mild-to-moderate chronic renal failure (S_Cr < 264 µmol/liter). Of the 748 cases of ARF studied, 665 episodes presented in inhabitans from the Madrid area. This gives an overall incidence of ARF of 209 cases per million population (p.m.p.; 95% CI 195 to 223). The incidence of acute tubular necrosis (ATN) was 88 cases p.m.p. (95% CI 79 to 97), prerenal ARF 46 p.m.p (95% CI 40 to 52), acute-onset chronic ARF 29 p.m.p. (95% CI 24 to 34), and obstructive ARF 23 p.m.p. (95% CI 19 to 27). The mean age was 63 ± 17 years. The most frequent causes of ARF were ATN (45%), prerenal (21%), acute-onset chronic renal failure (12.7%) and obstructive ARF (10%). Renal function was normal at admission in 48% of patients who later developed ARF. Mortality (45%) was much higher than that of the other patients admitted (5.4%, P < 0.001). This real outcome correlated extremely well with the expected outcome calculated through out the severity index of ARF (SI) 0.433 ± 0.246 (mean ± SD). In 187 cases, mortality was attributed to underlying disease, thus corrected mortality due to ARF was 26.7%. Dialysis was required in 36% of patients, and was associated with a significantly higher SI of ARF (0.57 ± 0.23 vs. 0.35 ± 0.19, P < 0.001) and mortality (65.9 vs. 33.2%, P < 0.001). Mortality in patients hemodialyzed with biocompatible synthetic membranes (N = 50) was similar to that observed with cellulosic ones (N = 84; 66% vs. 59.5%, NS). Mortality was higher in patients with coma, assisted respiration, hypotension, jaundice (all P < 0.001) and oliguria (P < 0.02). This study gives, for the first time, the incidence of all forms of ARF in a developed country. ARF is iatrogenically incuced at a high rate by modern medicine. Prevention strategies, Particularly in the perioperative period, are needed to decrease its impact.