Acute stroke commonly affects cardiac autonomic responses resulting in reduced heart rate variability (HRV). Multiscale entropy (MSE) is a novel non-linear method to quantify the complexity of HRV. This study investigated the influence of intracerebral hemorrhage (ICH) locations and intraventricular hemorrhage (IVH) on the complexity of HRV. We recruited 93 supratentorial ICH patients (male 59%, mean age 61 years), and the locations of ICH included basal ganglia (n = 40), thalamus (n = 35), and lobar (n = 18) group. Continuous one-hour electrocardiography signals were obtained from patients after admission, and the complexity index was defined as the area under the MSE curve. The complexity index was lower in lobar ICH (21.6 ± 7.9) than basal ganglia (27.9 ± 6.4) and thalamus (28.5 ± 7.2) groups. The complexity index was inversely correlated with initial stroke severity (r = -0.26), size of hematoma (r = -0.35) and ICH score (r = -0.26), especially among patients with intraventricular hemorrhage (r = -0.60, -0.60, and -0.41 respectively). A higher complexity can predict a good functional outcome (adjusted odds ratio 1.09, 95% confidence intervals 1.00-1.19) at post-stroke 3 months. In summary, more severe stroke and larger hematoma volume resulted in lower complexity of HRV. Lobar hemorrhage and IVH had great impacts on the cardiac autonomic function.