Background Since atezolizumab plus bevacizumab (ATE+BEV) regimen for patients with unresectable hepatocellular carcinoma (HCC) was released quite recently, real‐world data are lacking. We evaluated efficacy, safety, and predictive biomarkers for survival in patients receiving ATE+BEV. Methods Between 2020 and 2021, HCC patients receiving ATE+BEV at academic teaching hospitals were recruited. Treatment response was assessed using the Response Evaluation Criteria in Solid Tumors (version 1.1.). Results Among 121 patients enrolled, the median age was 63 years, with male predominance (82.6%). Complete response, partial response, stable disease, and progressive disease were identified in 2.5%, 26.4%, 54.5%, and 16.6%, respectively. Patients with alpha‐fetoprotein and des‐gamma‐carboxy prothrombin (DCP) response, defined as ≥30% and ≥50% decreases, respectively, at the first response evaluation relative to baseline, and those with neutrophil‐to‐lymphocyte ratio (NLR) <2.5, had significantly higher objective response rates (42.6% vs. 21.5%, 50.0% vs. 26.2%, and 39.0% vs. 19.4%, respectively; all p < 0.05). During follow‐up, the median overall survival (OS) was not reached, and the median progression‐free survival (PFS) was 5.7 months. Multivariable analyses showed that macrovascular invasion (adjusted hazard ratio aHR 2.541; p = 0.017), DCP ≥186 mAU/ml (aHR 5.102; p < 0.001), NLR ≥2.5 (aHR 3.584; p = 0.001), and an NLR decrease ≥10% at the first response (aHR 0.305; p = 0.002) were independent predictors of OS, and DCP ≥186 mAU (aHR 2.311; p = 0.002) and NLR ≥2.5 (aHR 1.938; p = 0.012) were independent predictors of PFS. Grade ≥3 treatment‐related adverse events (AEs) occurred in 33 (27.3%) patients. Conclusion ATE+BEV showed favorable efficacy and safety. Baseline high DCP and NLR may be useful prognostic predictors for OS and PFS.