OBJECT: In many new clinical trials of patients with malignant gliomas surgical intervention is incorporated as an integral part of tumor-directed interstitial therapies such as gene therapy, biodegradable wafer placement, and immunotherapy. Assessment of toxicity is a major component of evaluating these novel therapeutic interventions, but this must be done in light of known complication rates of craniotomy for tumor resection. Factors predicting neurological outcome would also be helpful for patient selection for surgically based clinical trials. METHODS: The Glioma Outcome Project is a prospectively compiled database containing information on 788 patients with malignant gliomas that captured clinical practice patterns and patient outcomes. Patients in this series who underwent their first or second craniotomy were analyzed separately for presenting symptoms, tumor and patient characteristics, and perioperative complications. Preoperative and intraoperative factors possibly related to neurological outcome were evaluated. There were 408 patients who underwent first craniotomies (C1 group) and 91 patients who underwent second ones (C2 group). Both groups had similar patient and tumor characteristics except for their median age (55 years in the C1 group compared with 50 years in the C2 group; p = 0.006). Headache was more common at presentation in the C1 group, whereas papilledema and an altered level of consciousness were more common at presentation in patients undergoing second surgeries. Perioperative complications occurred in 24% of patients in the C1 group and 33% of patients in the C2 group (p = 0.1). Most patients were the same or better neurologically after surgery, but more patients in the C2 group (18%) displayed a worsened neurological status than those in the C1 group (8%; p = 0.007). The Karnofsky Performance Scale score and, in patients in the C2 group, tumor size were important neurological outcome predictors. Regional complications occurred at similar rates in both groups. Systemic infections occurred more frequently in the C2 group (4.4 compared with 0%; p < 0.0001) as did depression (20 compared with 11%; p = 0.02). The perioperative mortality rate was 1.5% for the C1 group and 2.2% for the C2 group (p = not significant). The median length of the hospital stay was 4 days in each group. CONCLUSIONS: Perioperative complications occur slightly more often following a second craniotomy for malignant glioma than after the first craniotomy. This should be considered when evaluating toxicities from intraoperative local therapies requiring craniotomy. Nevertheless, most patients are neurologically stable or improved after either their first or second craniotomy. This data set may serve as a benchmark for neurosurgeons and others in a discussion of operative risks in patients with malignant gliomas.