Observations on the clinical effects of venesection therapy in 85 treated, as compared with 26 untreated, patients with idiopathic haemochromatosis showed decreased pigmentation and hepatomegaly together with a return to normal tests of liver function in half the patients who had abnormal tests at presentation. Control improved in 28 per cent of those patients with diabetes mellitus, although some patients developed it during the period of observation, despite venesection. Portal hypertension, testicular atrophy and arthropathy were not improved. In only 12 patients was there sufficient reaccumulation of iron after the initial course of venesection to merit further treatment. Rates of iron accumulation in these patients varied between 1-4 mg and 4-8 mg per day and chelatable iron levels were noted to be inappropriately high in relation to body iron stores during the early stages of the reaccumulation period. Life table data shows that the percentage survival five and ten years after diagnosis was 66 and 32 per cent respectively for the treated patients, and 18 and 6 per cent respectively for the untreated patients, both statistically highly significant differences (p less than 0-01). Possible clinical differences such as age of presentation, the presence of diabetes mellitus, cirrhosis, clinical hepatic failure and hepatoma between the treated and untreated groups that might otherwise have weighted survival in favour of the treated group were corrected by the use of covariant analysis. This gave mean log survival values of 4-15 and 2-88 for the treated and untreated patients respectively, equivalent to 63-4 months and 17-8 months, a highly significant difference (p less than 0-01). Ten patients, all of whom had cirrhosis at the time of diagnosis, died of malignant hepatoma between three and 15 years after completing venesection therapy. There was also a high rate of death from neoplasms in a variety of other sites--22 per cent in the venesected group, strikingly higher than that rate predicted for a similarly aged population using national cancer mortality rates.