The results of transsphenoidal adenomectomy for growth hormone (GH)-secreting pituitary adenomas in acromegaly performed over a 17-year period were analyzed retrospectively to determine which preoperative factors significantly influenced the long-term surgical outcome. These variables were then used to develop a logistic regression model to determine the probability of surgical failure. The series consisted of 103 patients. Long-term follow-up study (mean duration 102 +/- 64 months) was performed to derive outcome analysis and determinants of failure. Surgical control was defined as a long-term postoperative serum basal GH level of less than 5 micrograms/liter, a long-term postoperative serum somatomedin C (SM-C) level of less than 2.2 U/ml, and a favorable clinical response. Eighteen (17.5%) patients did not meet these criteria. The overall control rate by the GH criteria was 81.3% and by the SM-C criteria 76.2%. By multivariate logistic regression analysis, tumor stage was the strongest predictor of outcome (p < 0.05). The preoperative GH level, tumor grade, and preoperative SM-C level were significant univariate predictors (p < 0.05). There were statistically significant differences in mean preoperative GH and SM-C levels (p < 0.05, t-test) and tumor stage (p < 0.05, chi-squared test) between patients whose acromegaly was controlled by surgery and those whose acromegaly was not. Furthermore, estimates were derived of the probability of surgical failure based on preoperative GH level, preoperative SM-C level, and tumor grade and stage. The authors believe these findings will enhance clinical decision-making for neurosurgeons considering transsphenoidal microsurgery in patients with acromegaly.