The impact of midline shift (MLS) on long-term survival and progression in glioblastoma (GBM) is unknown. The objective of this study was to analyze the influence of mass effect on survival and progression with consideration of the patient demographics, tumor morphology, operative techniques, molecular pathology, and postoperative treatment. One hundred ninety-eight patients with GBM were analyzed retrospectively. Both MLS groups (< or ≥ 10 mm) were compared with regard to survival, progression-free survival (PFS), and postoperative course of Karnofsky Performance Status (KPS). A two-sided Fisher exact test showed no statistically significant differences in the confounders between the low- and high-MLS groups. The median survival was 18.0 months (95% confidence interval (CI) = 15.3-20.7) in the low-MLS group (n = 173) and 9.0 months (95% CI = 4.8-13.2) in the high-MLS group (n = 25) (p = 0.045). In the high-MLS group, 59.1% (13/22) with an initially high MLS had a KPS of less than 70% after 3 months, whereas 20.5% of the low-MLS group had a KPS of less than 70% (p < 0.001). Binary logistic regression analysis including the O-6-methylguanine-DNA methyltransferase (MGMT) status, extent of resection, baseline KPS, and MIB-I index showed low MLS as the only predictor for survival at 12 months (p = 0.046, odds ratio (OR) = 2.70, 95% CI = 1.0-7.2). Median PFS was 6.0 months in the high-MLS group and 9.0 months in the low-MLS group (log-rank test; p = 0.08). An initial midline shift of 10 mm or greater seems to be an imaging characteristic that independently predicts the survival in glioblastoma.