Cell typing is an important step in the single-cell RNA sequencing (scRNA-seq) analysis. Although some cell marker databases and cell typing tools have been proposed, limited roles are in prostate cell typing. Through literature review, we found prostate cell typing relied much on researchers' knowledge and experience, thus different markers were used to label the same cell type, leading to the divergences between studies, emphasizing the importance of a sound epistemological foundation for prostate cell typing in single-cell data. Therefore, we designed this study to provide performance analysis for prostate cell markers using eight integrated human prostate scRNA-seq datasets of 170,438 cells from 41 peoples (methods were described in Supplementary Data 2 in detail). Using unsupervised learning, information entropy, F1-score, and local outlier factor score, an objective performance analysis report was obtained, based on which, stable and specific human prostate main and fine cell markers were proposed. Our findings will help decide to select suitable markers for human prostate cell typing in single-cell data.