PATHOLOGICAL AND CLINICAL data from 56 patients operated on for craniopharyngioma since 1981 were analyzed to determine the utility of dividing patients with this tumor into distinct clinical groups based on recognized pathological type and to determine the prognostic import of brain invasion. Of the tumors in the 30 adult patients, 66% were adamantinomatous, 28% were squamous papillary, and the remainder were mixed. However, of the tumors in the 26 children, 96% were adamantinomatous and none were pure squamous papillary (P< 0.01). Forty-six percent of the children compared with 17% of the adults had brain invasion (P< 0.01). Brain invasion was present in 37% of the adamantinomatous but in only 13% of the squamous papillary tumors. Seventy-seven percent of the children underwent gross total resection (GTR) compared with 27% of the adults (P< 0.01). Sixty-three percent of the squamous papillary tumors underwent GTR compared with 54% of the adamantinomatous and mixed tumors. Follow-up ranged from 7 to 187 months (mean, 49 mo). After subtotal resection, with or without radiation therapy, 58% of the tumors recurred compared with 17% recurrence after GTR (P< 0.01), with a mean time to recurrence of 34 months. In both tumor histological types, subtotal resection was associated with a higher rate of tumor recurrence compared with gross total resection. Among the subtotally resected craniopharyngiomas, 2 of the 3 (67%) squamous papillary and 11 of the 21 (52%) adamantinomatous and mixed tumors recurred. In contrast, among the totally resected tumors, none of the 5 squamous papillary and only 5 of the 25 (20%) adamantinomatous and mixed tumors recurred. There were no significant differences in Karnofsky performance status score, mortality rate, or visual and endocrine outcomes when comparing patients based on histological tumor type. When controlling for age and extent of resection, we found that brain invasion had no significant effect on recurrence rate in totally resected tumors. Based on the limited number of patients in this series, we conclude as follows. 1) Contrary to previous reports, squamous papillary craniopharyngiomas, like adamantinomatous tumors, may recur when subtotally resected. 2) For both tumor variants, the most significant factor associated with craniopharyngioma recurrence is the extent of surgical resection rather than histopathological subtype. 3) Contrary to prior hypotheses, brain invasion in totally resected tumors does not predict higher recurrence. 4) GTR is associated with a significantly lower recurrence rate and can be achieved without sacrificing functional outcome.