Objective To investigate the role of three-dimensional (3D) reconstruction based parameters of hematoma cavity and encephalocoele in predicting hematoma expansion and hospitalized poor outcome in patients with primary brainstem hemorrhage (PBH). Methods Thirty-two PBH patients met research criterion were enrolled from intensive care unit (ICU) between June 2015 and December 2017. Baseline clinical characteristics, CT images on admission and within 48 h of admission were collected. The 3D reconstruction of hematoma cavity and encephalocoele based on CT images was performed by Mimics10.0, and quantity of triangles per square milimet surface (TQOT/mm2), and hematoma volume (HV) and encephalocoele volume (EV) were obtained. All patients were divided into hematoma expansion group and non-hematoma expansion group according to whether hematoma expansion appeared (hematoma expanded>33％ within 48 h of admission as compared with that on admission), and hospitalized poor outcome group and hospitalized non-poor outcome group according to whether hospitalized poor outcome appeared (modified Rankin scale scores>4 at discharge or hospitalized deaths), respectively. The risk factors of hematoma expansion were investigated by multivariable Logistic regression analysis. Multivariable Cox hazard regression was used to analyze the risk factors of poor outcome; Kaplain-Meier survival curve analysis and Log-rank test were used to compare the differences in survival curves between independent risk factors screened by Cox regression analysis. Results There were 11 patients (34.4％) with hematoma expansion and 14 (43.8％) with ventriculomegaly in 32 patients; in these 11 patients with hematoma expansion, 8 had ventriculomegaly, and the two had positive correlation (rp=0.423, P=0.016). Fifteen patients (46.9％) had poor outcome, in which 11 (34.4％) died in hospital; 5 had hematoma expansion and 8 had ventriculomegaly. Multivariate Logistic regression analysis showed that baseline lactate >2.0 mmol/L (OR=11.986, 95％CI: 1.084-132.552, P=0.043) and TQOT/mm2>2 (OR=10.223, 95％CI: 1.424-73.396, P=0.021) were independent risk factors of hematoma expansion. Baseline HV (HR=1.102, 95％ CI: 1.020-1.143, P=0.002) and EV (HR=3.485, 95％ CI:1.071-11.463, P=0.040) were risk factors of hospitalized poor outcome identified by multivariable Cox analysis. Kaplan-Meier survival analysis showed that the hospitalization days of hospitalized poor outcome were (74.0±10.6) d and (25.5±7.0) d between patients have hematoma expansion Cut-off value of 7 mL, with significant difference (Log-rank: χ2=11.832, P=0.001), and the hospitalization days of hospitalized poor outcome in patients with and without ventriculomegaly were (68.1±9.0) d and (29.9± 8.8) d, respectively, with significant difference (Log-rank: χ2=7.483, P=0.006). Conclusions There is correlation between hematoma expansion and ventriculomegaly; patients with TQOT/mm2>2 might have high risk of hematoma expansion; patients with baseline HV>7 mL and ventriculomegaly would sooner have hospitalized poor outcome.