We have used in vivo rectal dialysis to test the hypothesis that the diarrhoea of patients with active ulcerative colitis (UC) is due to inhibition of large intestinal salt and water absorption by enhanced local mucosal prostaglandin (PG) synthesis. In 28 patients with untreated UC, increased rectal mucosal PGE2 release varied inversely with sodium transport and directly with potassium transport; higher PGE2 release was also associated with lower (i.e. less negative) potential difference (PD). Disease activity assessed sigmoidoscopically was positively related to PGE2 release, potassium transport and lower PD. Similar relationships were found in 33 patients treated with sulphasalazine and/or corticosteroids and in 9 patients studied serially when on unaltered conventional, or no treatment. In contrast, when 10 patients with UC were given the PG synthesis inhibitor, flurbiprofen, a significant fall in PGE2 release was associated with deteriorations in mucosal PD and transport of sodium and potassium. Increased PG production is therefore unlikely to be a major determinant of the abnormalities of electrolyte transport found in UC. The correlations between mucosal PGE2 release and electrolyte transport were probably due to the relationship of these variables with some other consequence of tissue damage.