Percutaneous endoscopic gastrostomy (PEG) is the preferred route for long-term enteral feeding. Our aims were to prospectively evaluate the outcome ("PEG status") and complications of PEG and to determine whether these can be predicted by patients' baseline characteristics. We conducted a prospective study in two tertiary hospitals between August 2003 and January 2005, enrolling all patients who were undergoing PEG placement. We completed a questionnaire with details of demographic data, diagnosis, indication for PEG, Charlson's co-morbidity index, Barthel's index, laboratory tests, complications, and date and cause of death. Patients were followed at scheduled appointments. Univariate and multivariate analyses were performed. 168 patients (48% male, 52% female; mean age +/- standard deviation 74 +/- 16 years) underwent PEG using the pull technique. The main indication was neurogenic dysphagia (156 patients, 92.9%). Although most indications were appropriate, in half the cases these were established too late. There were no procedure-related deaths. Major complications occurred in four patients (2.4%); minor complications occurred in 52 patients (31%). No single variable could predict complications. Fifteen patients (9%) had the PEG removed. No single variable was independently associated with PEG removal. The mortality was 6.5% at 30 days, 17.3% at 90 days and 33.9% at 1 year. The C-reactive protein was the only predictive factor of early mortality (< or = 30 days), and Charlson's co-morbidity index was the only predictive factor of late mortality (> 30 days). PEG placement is an easy and safe procedure, although it is often requested too late. No single variable could predict complications or PEG removal. C-reactive protein was found to be predictive of early mortality and Charlson's index was predictive of late mortality.