[en] Purpose: To correlate the pattern of failure and subsequent survival with the technical accuracy of stereotactic brain tumor implantation. Methods and Materials: The patterns of failure of 47 consecutive patients with primary or recurrent gliomas after stereotactic implantation delivering 60 Gy via removable high activity ¹²⁵I sources were reviewed. When the tumor was covered at all levels by the chosen isodose distribution, the implant was considered to be 'ideal'. If the coverage was not complete, a numerical description of the volume of tumor outside the isodose was assigned. Criteria for 'adequate' and 'inadequate' implants were defined. Standard radiographic criteria, with pathologic confirmation in 26 cases, were used to categorize the patterns of failure into the following components: central, peripheral, distant (within the brain parenchyma), leptomeningeal, and spinal. A peripheral failure was scored as being in the 'direction of error' when the prescribed isodose did not cover the tumor volume and the subsequent tumor progression was in this region. Survival was calculated from the date of implantation. Results: Of 47 cases examined, 72% had an element of central and/or peripheral failure and 23% had a component of distant or meningeal failure. Among the patients with 'adequate' or 'inadequate' ('nonideal') implants who had a component of peripheral failure, only 19% were in the 'direction of error'. All patients with technically 'inadequate' implants progressed in both the central and peripheral region. Among the groups who had 'ideal', 'adequate', and 'inadequate' implants; 37%, 70%, and 75%, respectively, underwent reoperation [p = not significant (NS)]. Patients who underwent reoperation had a longer median survival than those who did not; 521 days vs. 298 days, respectively (p = 0.035). For patients with 'nonideal' implants, a median survival of 470 days was found for patients undergoing reoperation vs. 184 days for those who did not (p = 0.016). Conclusions: (a) Patients with 'inadequate' implants failed in both the central and peripheral region in all cases. This pattern, while less common in those with 'ideal' or 'adequate' implants, occurred in the majority of cases. (b) The technical excellence of the implant had no impact on survival. (c) Patients with 'nonideal' implants were more likely to have reoperation than those with 'ideal' implants, and this intervention was associated with a significant survival advantage