The molecular classification of breast cancer mainly focuses on ER, PR, and HER2 status detected by immunohistochemistry (IHC) analysis. To explore the clinical value of breast cancer classification based on gene-based diagnosis of the triple markers, we measured ESR1, PGR, and ERBB2 mRNA levels in 294 breast cancer patients by reverse transcription quantitative polymerase chain reaction (RT-QPCR), and examined their correlation with ER, PR, and HER2 status detected by IHC. We observed a significant positive correlation between the mRNA levels of the triple markers and their protein status (ESR1 vs. ER, Spearman's ρ = 0.527, P = 2.3 × 10(-22); PGR vs. PR, Spearman's ρ = 0.631, P = 5.1 × 10(-34); ERBB2 vs. HER2, Spearman's ρ = 0.439, P = 3.0 × 10(-15)). Furthermore, the subtypes determined by mRNA levels of the triple markers were significantly correlated to the subtypes determined based on their protein status (Spearman's ρ = 0.342, P = 2.0 × 10(-8)). Kaplan-Meier analysis showed that the subtypes determined by mRNA levels of the triple-marker could predict the disease-free survival (DFS) in breast cancer patients. Multivariate analysis showed that the predictive value of DFS could be confirmed for the subtypes determined by mRNA levels of the triple markers (HR = 2.285, P = 0.008) but not for those determined by their protein status. Taken together, our results suggest that the detection of ESR1/PGR/ERBB2 mRNA levels by RT-QPCR is a better approach for subtyping breast cancer and predicting the prognosis.