A study was conducted to see whether patients with duodenal ulcers that failed to heal in response to H2 receptor antagonists had a higher incidence of recurrent ulceration after highly selective vagotomy than patients whose ulcers healed with these drugs. Between 1977 and 1983, 157 patients had a highly selective vagotomy for uncomplicated duodenal ulcer; in 57 patients the ulcer had failed to heal despite treatment with H2 receptor antagonists (refractory group), 19 patients had developed recurrent ulceration while receiving maintenance treatment, 67 patients had remained healed while taking H2 receptor antagonists but suffered frequent relapses when treatment was stopped, and 14 patients had not been given these drugs before operation. The overall incidence of recurrent ulceration was 6% after two years and 11% after five years of follow up. In the refractory group, however, the incidence of recurrent ulceration was 18% at two years and 34% after five years, whereas the incidence of recurrence was only 1.5% at two years and 3% after five years in patients whose ulcers had healed with H2 receptor antagonists. Resistance to H2 receptor antagonists was not related to preoperative basal or peak acid output but was related to cigarette smoking. Factors associated with recurrent ulceration after highly selective vagotomy were basal acid outputs before and after operation, cigarette smoking, and the surgeon who performed the operation. Duodenal ulcers that fail to respond to H2 receptor antagonists represent a more severe ulcer diathesis, for which highly selective vagotomy is less effective.