BACKGROUND Adjuvant treatment for node-negative breast cancer remains controversial. It is important to pick out the high-risk groups who may benefit from adjuvant systemic therapy and avoid the unnecessary additional therapy for the favorable prognostic groups. METHODS Retrospective study of immunohistochemical staining for HLA-DR on tumor cells from paraffin-embedded tissue specimens of 32 patients with node-negative invasive breast cancer at this hospital from 1986 to 1991 was performed with the aim to investigate its prognostic significance. RESULTS HLA-DR staining was positive in nine (28%) patients and negative in twenty-three (72%) patients. One (11%) was a recurrence, and no (0%) death occurred in the positive group, compared with nine (39%) recurrences and five (22%) deaths in the negative group. The multivariate analysis failed to show that HLA-DR expression is an independent prognostic factor. However, with univariate analysis, the 5-year disease-free survival rate (87%) of the positive group was significantly better than that (35%) of the negative group (p = 0.04). The 5-year overall survival rate (100%) of the positive group was also better than that (66%) of the negative group, but the difference was not statistically significant. Furthermore, when combination of the HLA-DR expression and estrogen receptor status was used, both the 5-year disease-free and overall survival rate (81% and 100%, respectively) of group A (positive staining for either HLA-DR or estrogen receptor and positive staining for both) were significantly better than those (33% and 58%, respectively) of group B (negative staining for both HLA-DR and estrogen receptor). CONCLUSIONS We believe HLA-DR expression may be a promising, additive predictive factor to node-negative breast cancer and deserves further investigation based on these preliminary results.