BACKGROUND: This study investigated the hypothesis that separate phenotypes of Crohn's disease exist which display differing patterns of recurrence with a tendency to preservation of phenotype between serial operations. METHODS: Some 483 abdominal operations (278 patients) were identified from a prospectively compiled database. Patterns of recurrence (reoperation) were analysed by Kaplan-Meier plots and log rank tests according to disease phenotype (perforated, stenosed or ulcerated). Serial operations were analysed by agreement of phenotype and microscopic features of disease using kappa statistics and correlation coefficients. RESULTS: There was no significant difference in recurrence according to disease phenotype (median reoperation-free survival time 43.0, 50.2 and 47.9 months for perforated, stenosed and ulcerated types respectively; log rank chi(2) = 3.5, P = 0.18). There was poor agreement in phenotype between serial operations (kappa = 0.22 for first/second operation and kappa= 0.15 for second/third operation) and no significant correlation between pathological features was identified (r between -0.19 and 0.48). CONCLUSION: No evidence was found for the existence of separate disease phenotypes with differing natural histories or underlying pathological characteristics.