To establish whether concentrations of eicosanoids determined by equilibrium in vivo dialysis of faeces and equilibrium in vivo dialysis of rectum might predict a relapse in ulcerative colitis, 23 patients with completely inactive disease, maintained on sulphasalazine, stopped treatment and entered a prospective study. Concentrations of prostaglandin E2 were determined by radioimmunoassay on purified faecal and rectal dialysates at entry, at two weeks, and at two, six, and 12 months. If the above concentrations exceeded control concentrations (0.5 ng/ml and 1.0 ng/ml in faecal and rectal fluid, respectively) at any study day, the patient was allocated at random to double blind treatment with sulphasalazine 2 g/day, or placebo for six months. A relapse, defined as recurrence of symptoms accompanied by endoscopic inflammation occurred in none of six and in four of five patients allocated to sulphasalazine and placebo, respectively (p less than 0.05). In no case a normal rectal prostaglandin E2 concentration was associated with a relapse in the short term, but only two of 12 patients observed passively remained in remission. In retrospect, leukotriene B4 was a less sensitive predictor of relapse than prostaglandin E2. We conclude that raised concentrations of prostaglandin E2 in rectal dialysis fluid identify patients with a substantial risk of relapse.