Eighty-five medically intractable trigeminal neuralgia patients treated by percutaneous retrogasserian glycerol rhizotomy (PRGR) were followed for 6 to 54 months. The median time to recurrence of symptoms refractory to medical therapy and requiring further intervention was 3 years (by Kaplan-Meier survival analysis). The median time to recurrence of symptoms requiring some form of medical treatment was 2 years. Following repeat PRGR for recurrent symptoms, the median time to recurrence was 1 year. Univariate log rank statistics and multivariate Cox proportional hazards modeling revealed significant associations between favorable outcome and female sex, absence of atypical features or associated cluster headache symptoms, success of prior carbamazepine therapy, duration of symptoms, and cerebrospinal fluid return during the procedure. A scoring system using prognostic factors has been developed based upon these findings. Assessment of published studies of PRGR and of other treatments for trigeminal neuralgia is made difficult by the variety of outcome measures employed and variable follow-up intervals. The present study attempts to address these issues by definition of endpoints, statistical analysis of the data, and identification of important prognostic factors in a manner useful to the clinician.