Background Although the diagnostic value of plasma heat-shock protein 90 alpha (HSP90 alpha) in hepatocellular carcinoma (HCC) has been previously reported, the causal effect of the plasma HSP90 alpha levels on HCC prognosis remains largely unclear. To this extent, we sought to assess whether the plasma HSP90 alpha acts as a prognostic factor for HCC patients. Methods A total of 2150 HCC patients were included in this retrospective study between August 2016 and July 2021. Plasma HSP90 alpha levels were tested within a week before treatment and their association with prognosis was assessed. Results An optimal cutoff value of 143.5 for the HSP90 alpha based on the overall survival (OS) was determined using the X-tile software. HCC patients with HSP90 alpha < 143.5 ng/mL (low HSP90 alpha) before and after propensity score matching (PSM) indicated longer median OS (mOS) relative to those with HSP90 alpha >= 143.5 ng/mL (high HSP90 alpha) (37.0 vs. 9.0 months, p < 0.001; 19.2 vs. 9.6 months, p < 0.001; respectively). In addition, the high HSP90 alpha plasma level is an independent poor prognostic factor for OS in HCC patients. In our subgroup analysis, including the supportive care group, surgery group, transarterial chemoembolization (TACE) group, adjuvant TACE group, an immune checkpoint inhibitor (ICI) plus targeted therapy group, and TACE plus ICI group, the high HSP90 alpha group demonstrated better OS compared to the low HSP90 alpha group. Moreover, in the supportive care, TACE, ICI plus targeted therapy, TACE plus ICI groups, and high HSP90 alpha levels were also an independent poor prognostic factors for OS. Conclusions Our study confirmed that the plasma HSP90 alpha level can be used as a prognostic biomarker for HCC.