The aim of the present prospective study was to determine if malnutrition, measured using a simple validated score, is an independent risk factor for nosocomial infections (NI) in non-selected hospital in-patients. Between 29 and 31 May 2001, a survey on the prevalence of NI was conducted on all 1637 in-patients (61 (sd 25) years old) in a French university hospital as part of a national survey. Actual and usual body weights were recorded in all in-patients, and serum albumin levels were measured on all blood samples taken during the week before the study. Nutritional status was evaluated by using the nutritional risk index (NRI). Albumin values were obtained in 1084 patients, and complete weight information was obtained in 911. Therefore, NRI was calculated in 630 patients (61 (sd 20) years old): 427 (67·8%) were malnourished. NI prevalence was 8·7%: 4·4% in non-malnourished patients, 7·6% in moderately malnourished patients and 14·6% in severely malnourished patients. In univariate analysis, the odds ratios for NI were 1·46 (95% CI 1·2, 2·1) in moderately malnourished patients and 4·8 (95% CI 4·6, 6·4) in severely malnourished patients. In multivariate analysis, age, immunodeficiency and NRI class influenced NI risk. Vascular and urinary catheters, and surgical intervention, were the extrinsic factors associated with NI, with odds ratios ranging from 2·0 (95% CI 1·8, 2·6) for vascular catheters to 10·8 (95% CI 8·8, 12·6) for association of the three factors. In conclusion, in non-selected hospitalized patients, malnutrition assessed with a simple and objective marker is an independent risk factor for NI. An early screening for malnutrition may therefore be helpful to reduce the high prevalence of NI.