Background: most studies that analyze the influence of structure factors on clinical outcomes are retrospective, based on clinical-administrative databases, and mainly focusing on surgical volume. Objective: to study variations in the process and outcomes of oncologic surgery for esophagus, stomach, pancreas, liver metastases and rectum cancers in Catalonia, as well as the factors associated with these variations. Patients and method: a retrospective (2002) and prospective (2003-05) multicenter cohort study. Data forms were designed to collect patient, process, and care outcome characteristics before surgery, at hospital discharge, and at 3 and 6 months after discharge. Main outcome measures were hospital and followup mortality, complications, re-interventions, and relapse rates. Results: 49 hospitals (80%) participated in the retrospective phase, 44 of which (90%) also participated in the prospective phase: 3,038 patients (98%) were included. No differences were observed in the profile of operated patients according to hospital level of complexity, but clinical-pathological staging and other functional status variables could not be assessed because of over 20% of missing values. There was significant variability in the volume of interventions as well as in certain aspects of the healthcare process depending on type of cancer and center complexity. High rates of esophageal cancer mortality (18.2% at discharge, 27.3% at 6 months) and of complications and re-interventions for all cancers assessed, especially rectal cancer (18.4% re-interventions at 6 months), were identified. Conclusions: the study of the variability identified will require adequate risk-adjustment and should take into account different structure factors. It is necessary that information included in medical records be improved.