Oral tobramycin for 7 days has been shown to be of benefit as an adjunct to conventional medication in acute ulcerative colitis. Eighty-one patients (40 who had received tobramycin; 41 placebo) who had been enrolled in a double-blind placebo-controlled trial of this drug in acute disease were subsequently followed to determine whether this short-term benefit persisted. Relapse was defined as a liquid stool frequency of three times daily with rectal bleeding. Results were analysed by the log-rank test on Kaplan-Meier survival curves. Treatment failure was defined as a lack of response by the end of the acute trial period, or subsequent relapse. In a second analysis, only those entering remission at the end of the acute trial were considered, and followed to relapse. Although at the start of the follow-up period significantly fewer patients in the tobramycin group had failed (failed: tobramycin 9, placebo 24; not failed tobramycin 31; placebo 17; P = 0.001), the failure-free survival curves subsequently converged and did not differ significantly. After 1 and 2 years, the failure-free survival rates were 40% (S.E. = 7.8%) and 20% (S.E. = 6.3%) for the tobramycin group and 24% (S.E. = 6.7%) and 12% (S.E. = 5.1%) for the placebo group. When only those entering remission were considered, there was no significant difference in the relapse rates in the two groups. Benefit from tobramycin is therefore short-lived and may reflect short-term changes in the faecal flora.