The effect of Yag laser photocoagulation on the course of bleeding of gastrointestinal vascular malformations was studied in 59 patients, with a total of 482 lesions. The lesions were located in the upper gastrointestinal tract alone in 25 patients, in the lower tract alone in 31 patients and in both the lower and the upper gastrointestinal tract in three patients. In the month before laser therapy the number of bleeding episodes averaged 1.09 +/- 0.6 (SD) per patient (n = 57) and the transfusion requirements 2.4 +/- 2.6 red blood cells units per patient, while in the month after treatment the bleeding incidence averaged 0.16 +/- 0.5 and the transfusion requirements 0.21 +/- 0.8 (both p less than 0.001). Long term results were analysed considering for each patient an equally long pretreatment and follow up period. After a mean follow up period of 11.5 months (1-48 months), 17 of the 57 patients available for follow up rebled. The reduction of the bleeding rate was statistically significant at one, six, 12, and 18 months of follow up, while transfusion rate was significantly decreased at one, six, and 12 months. The results were disappointing in patients with Osler-Weber-Rendu (n = 4) and in patients with angiomas associated with Von Willebrand's disease (n = 3), who all rebled. In angiodysplasia the treatment was successful in 82% of the 49 patients. The more numerous the lesions, the less effective the reduction in bleeding rate by laser treatment was. Histological studies showed that the haemostatic effect of Yag laser photocoagulation was obtained by destruction of the lesion. Rebleeding was due to lesions missed at the first treatment, incompletely treated lesions and recurrence of new lesions. In two patients a free caecal perforation necessitated a right hemicolectomy. In both patients numerous or very large lesions had been treated in the caecum.