The use of frozen section examination of resection margins in selecting the site for an ileocolonic anastomosis was evaluated in patients with Crohn's disease. Seventy-nine patients with ileitis or ileocolitis who underwent first resection with removal of all apparent disease and primary ileocolonic anastomosis were studied. The patients with a resection margin examined by frozen section (FS group, n = 38) and those with margins chosen on the basis of visual inspection alone (NO-FS group, n = 41) were compared. Frozen section examination was found to be poor at detecting margin involvement: although 60 of 61 margins examined by frozen section were reported as negative at the time of operation, 20 of 61 were actually involved by Crohn's disease. Furthermore, despite the use of frozen section examination, the prevalence and severity of margin involvement in the FS and NO-FS groups were not statistically significantly different. Short term and long term clinical outcome were also not statistically different; the incidence of postoperative anastomotic leakage and obstruction was 13 per cent in the FS group and 5 per cent in the NO-FS groups (p NS). Clinical recrudescence rates by life table analysis in the FS and NO-FS groups were 37 +/- 9 per cent and 50 +/- 8 per cent, respectively, at five years, and 60 +/- 12 per cent and 66 +/- 9 per cent at ten years (p NS). Reoperation rates were 18 +/- 8 per cent at five years in both groups and 36 +/- 13 per cent in the FS group and 32 +/- 12 per cent in the NO-FS group at ten years. The findings support conservative resection to achieve grossly uninvolved margins rather than sacrifice of functional intestine in an attempt to achieve histopathologically uninvolved margins on frozen section.