Studies of acromegaly have shown a doubling of mortality compared with the general population. With the development of new modalities of treatment, it has become important to identify prognostic factors relating to mortality. Between 1964 and 2000, 208 acromegalic patients were followed for a mean of 13 yr at Auckland Hospital. Treatment was by surgery or radionuclide pituitary implantation, and all except 27 patients received pituitary radiation. Over the duration of the study, 72 patients died at a mean age of 61 +/- 12.8 yr. Those dying were significantly older at diagnosis, had a higher prevalence of hypertension and diabetes, and were more likely to have hypopituitarism. The observed to expected mortality ratio (O/E ratio) fell from 2.6 (95% confidence interval, 1.9-3.6) in those with last follow-up GH greater than 5 microg/liter to 2.5 (1.6-3.8), 1.6 (0.9-3), and 1.1 (0.5-2.1) for those with GH less than 5, less than 2, and less than 1 microg/liter, respectively (P < 0.001). Serum IGF-I, expressed as an SD score, was significantly associated with mortality, with O/E mortality ratios of 3.5 (95% confidence interval, 2.8-4.2) for those with an SD score greater than 2, 1.6 (0.6-2.6) for those with an SD score less than 2 (normal or low levels), and 1.0 (0.1-3) for those with an SD score less than zero. When assessed by multivariate analysis, last serum GH (P < 0.001), age, duration of symptoms before diagnosis (P < 0.03), and hypertension (P < 0.04) were independent predictors of survival. If IGF-I was substituted for GH, then survival was independently related to last IGF-I SD score (P < 0.02), indicating that GH and IGF-I act equivalently as predictors of mortality. These findings indicate that reduction of GH to less than 1 microg/liter or normalization of serum IGF-I reduces mortality to expected levels.