This study aimed to comprehensively evaluate the diagnostic performance of dual-energy CT (DECT) for differentiating acute intracranial hemorrhage (ICH) from contrast staining or small calcifications via a systematic review and meta-analysis. The PubMed-MEDLINE, EMBASE, and Cochrane Library databases were searched up to November 10, 2019. Original studies (prospective or retrospective cohort studies) with the primary aim of detecting ICH using DECT were selected. The diagnostic performance of DECT was assessed using bivariate and hierarchical summary receiver operating characteristic models. Quality assessment was performed according to the Quality Assessment of Diagnostic Accuracy Studies-2, while between-study heterogeneity was assessed using Higgins' inconsistency index (I2). To explore heterogeneity, subgroup meta-regression analyses were performed. Deeks' funnel plot asymmetry test was used for assessing publication bias. Nine studies comprising 402 patients with 453 lesions were included for data synthesis. The overall pooled sensitivity and specificity of DECT for ICH detection were 96% (95% CI, 77-99%) and 98% (95 CI, 93%-100%), respectively. Substantial and moderate between-study heterogeneities were observed for sensitivity (I2 = 90.3%) and specificity (I2 = 57.9%), respectively. In meta-regression analysis, type of cohort affected heterogeneity-studies including only stroke patients showed lower sensitivity (43.5% vs. 94.2%) but higher specificity (98.7% vs. 92.6%) than those with mixed etiologies (P < 0.001). Deeks' funnel plot asymmetry test revealed publication bias (P = 0.020). DECT demonstrated excellent diagnostic performance in terms of differentiating acute ICH from contrast staining and small calcifications. However, publication bias suggests the possibility of overestimated diagnostic performance, warranting large-scale, prospective cohort studies.