The complications of emergency surgery for acute upper gastro-intestinal haemorrhage in 95 patients were compared with those of surgery for non-bleeding benign peptic ulceration in 380 patients. There was a significantly higher number of senior surgeons performing emergency surgery for haemorrhage. There was a significantly higher mortality (10-5 %), wound dehiscence rate (12-6 %) and wound complication rate with re-bleeding (33-7 %) in the post-haemorrhage patients compared with the non-bleeding group (1-3, 1-8 and 11-6 % respectively). There was no significant difference in the incidence of chest infection in the two groups. Mortality and morbidity after haemorrhage were correlated directly with the amount of blood transfused before surgery. Rats were used to study the effect of the removal of 13 % of the blood volume, and its restoration after varying time intervals, upon subsequent wound healing. Abdominal wounds were tested at 7 days for breaking and bursting strength. There was no difference between the wound strength of the control animals and animals which had been bled for 4 minutes before wounding. A significant reduction of wound strength occurred in animals which had been bled for 30-60 minutes before wounding. Prolonged blood loss was associated with deficient wound healing even if blood loss was restored before the wounds were made. Stored blood may be deficient in a factor which is necessary for normal wound healing.