Background: The role of surgery for recurrent glioblastoma is under debate. Tumor volume at recurrence was prognostic in single-institution retrospective studies, but no data from prospective multicenter trials for recurrent glioblastoma have been published to date. Here, we report the association with outcome of surgery for recurrent glioblastoma in a well characterized patient cohort treated in the DIRECTOR trial, which evaluated the effect of two different dose-intensified temozolomide regimens at first recurrence of glioblastoma. Methods: We analyzed prospectively collected clinical, molecular and imaging data from the DIRECTOR cohort (N = 105). Imaging data was available from 87 patients. Volumetric analysis was performed based on gadolinium (Gd) enhancement on MRI and correlated with progression-free survival (PFS) and overall survival (OS). Proportional hazard models were applied to obtain prognostic factors. Results: 71 of 105 patients received surgery at recurrence. Prognostic factors were balanced between patients who had undergone surgery and those who had not, including age (P = 0.358), O⁶-methylguanine DNA methyltransferase (MGMT) gene promoter methylation (P = 0.965), isocitrate dehydrogenase (IDH)-1 mutation (P = 0.724), Karnofsky performance score (P = 0.880), or steroid intake before randomization (P = 0.950). Mean tumor volumes in patients who had received surgery were smaller at study entry than in patients who had not undergone surgery (3.0 cm³ [range 0-37] versus 6.8 cm³[range 1-23], P < 0.001). The outcomes in patients who did or did not receive surgery at recurrence were similar for PFS (2.0 months [95% CI 3.5-7.1] vs. 1.9 months [95% CI 1.9-6.4], P = 0.1974) and OS (9.2 months [95% CI 8.7-12.6] vs 9.4 months [95% CI 8.3-13.5], P = 0.9538). Among patients who underwent surgery for recurrence, post-surgery imaging was available in 59 cases. In these patients, complete resection of Gd-enhancing tumor (N = 39) versus residual detection of Gd enhancement (N = 20) was associated with improved OS (11.5 months [95% CI 9.3-15.1] vs. 6.7 months [95% CI 5.2-9.5], P = 0.006). Conclusions: Surgery at first recurrence of glioblastoma may improve outcome if complete resection of Gd-enhancing tumor is feasible.