Introduction: This study investigated the relationship between nutritional status, as determined by the prognostic nutritional index (PNI), and outcomes in patients with unresectable hepatocellular carcinoma (HCC) treated with atezolizumab plus bevacizumab (Atez/bev). Methods: The study analyzed 485 HCC patients treated with Atez/bev. Results: There were 342 patients with a low PNI (<47) and 143 patients with a high PNI (≥47). The median follow-up duration was 9.4 (6.0–14.3) months. Multivariate Cox hazards analysis showed that an α-fetoprotein level ≥100 ng/mL (hazard ratio (HR), 2.217; 95% confidence interval (CI), 1.588–3.095; p < 0.001), and PNI ≥47 (HR, 0.333; 95% CI, 0.212–0.525; p < 0.001) were independently associated with overall survival. Multivariate analysis showed that an α-fetoprotein level ≥100 ng/mL (HR, 1.690; 95% CI, 1.316–2.170; p < 0.001) and PNI ≥47 (HR, 0.696; 95% CI, 0.528–0.918; p = 0.010) were independently associated with progression-free survival. Cumulative overall and progression-free survival rates differed significantly by PNI ( p < 0.001 and p < 0.002, respectively). In a subgroup analysis using inverse probability weighting adjustment in patients with albumin-bilirubin grade 1 ( n = 173), univariate Cox hazards analysis showed that a PNI ≥47 (HR, 0.502; 95% CI, 0.260–0.991; p = 0.047) was significantly associated with overall survival. Spline curve analysis revealed that a PNI of approximately 34–48 is an appropriate cutoff for predicting good overall and progression-free survival. Conclusion: The PNI, a biomarker of nutritional status, can predict prognosis in patients with HCC treated with Atez/bev, even those who are considered to have a good prognosis due to good liver function.