OBJECTIVE Decompressive craniectomy (DC) is an established part of treatment in patients suffering from malignant infarction of the middle cerebral artery (MCA) or traumatic brain injury (TBI). However, no clear evidence for intracranial pressure (ICP)-guided therapy after DC exists. The lack of this evidence might be due to the frequently used, but simplified threshold for ICP of 20 mm Hg, which determines further therapy. Therefore, the objective of this study was to evaluate this threshold's accuracy and to investigate the course of ICP values with respect to neurological outcome. METHODS Data on clinical characteristics and parameters of the ICP course on the intensive care unit were collected retrospectively in 102 patients who underwent DC between December 2007 and April 2014 at the authors' institution. The postoperative ICP course in the first 168 hours was recorded and analyzed. From these findings, ICP thresholds discriminating favorable from unfavorable outcome were calculated using conditional inference tree analysis. Additionally, survival analysis was performed using the Kaplan-Meier method. Prognostic factors were assessed via univariate analysis and multivariate logistic regression. Favorable outcome was defined as a score of 0-4 on the modified Rankin Scale. RESULTS Multivariate logistic regression revealed that anisocoria, diagnosis, and ICP values differed significantly between the outcome groups. ICP values in the favorable and unfavorable outcome groups differed significantly (p < 0.001), while the mean ICP of both groups lay below the limit of 20 mm Hg (17.5 and 11.5 mm Hg, respectively). These findings were reproduced when analyzing the underlying pathologies of TBI and MCA infarction separately. Based on these findings, optimized time-dependent threshold values were calculated and found to be between 10 and 17 mm Hg. These values significantly distinguished favorable from unfavorable outcome and predicted 30-day mortality (p < 0.001). CONCLUSIONS This study systematically evaluated ICP levels in a long-term analysis after DC and provides new, surprisingly low, time-dependent ICP thresholds for these patients. Future trials investigating the benefit of ICP-guided therapy should take these thresholds into consideration and validate them in further patient cohorts.