Aims: To assess the impact of a simple computer-based decision-support system (computer help) on the quality of nutrition support orders and patients' outcome in Intensive-Care Unit (ICU). Methods: This quality-improvement study was carried out in a 16-bed medical-surgical ICU in a French university hospital. All consecutive patients who stayed in ICU more than 10 days with non-oral feeding for more than 5 days were retrospectively included during two 12-month periods. Prescriptions of nutrition support were collected and compared to French national guidelines as a quality-improvement process. A computer help was constructed using a simple Excel-sheet (MicrosoftTM) to guide physicians' prescriptions according to guidelines. This computer help was displayed in computers previously used for medical orders. Physicians were informed but no systematic protocol was implemented. Patients included during the first (control group) and second period (computer help group) were compared for achievement of nutrition goals and ICU outcomes. Results: The control and computer help groups respectively included 71 and 95 patients. Patients' characteristics were not significantly different between groups. In the computer help group, prescriptions achieved significantly more often 80% of nutrition goals for calorie (45% vs. 79% p<0.001) and nitrogen intake (3% vs. 37%, p<0.001). Incidence of nosocomial infections decreased significantly between the two groups (59% vs. 41%, p = 0.03). Mortality did not significantly differ between control (21%) and computer help groups (15%, p = 0.30). Conclusions: Use of a widespread inexpensive computer help is associated with significant improvements in nutrition support orders and decreased nosocomial infections in ICU patients. This computer-help is provided in electronic supplement.