BACKGROUND: Prognostic factors for predicting the recurrence of node-negativebreast cancers have been controversial. The present study was performed to elucidate practically useful prognostic factors using formalin-fixed paraffin sections. METHODS: This was a case-controlled multi-institutional study that composed 40 patients with recurrent node-negative breast cancer and 80 patients with node-negative breast cancer but without recurrence after radical surgery. Tumors weresmaller than 3 cm in diameter and were treated surgically between January 1, 1985 and December 31, 1990. The recurrent and non-recurrent cases were matched with regard to their age, adjuvant chemotherapy and the year in which surgery was performed. Fourteen immunohistochemical factors and 8 histological factors of theprimary tumor were studied on formalin-fixed, paraffin-embedded sections by immunohistochemical and histochemical analyses. RESULTS: According to univariate analysis, factors such as progesterone receptor (PgR), MIB-1, CD44v6, CD44v9 and platelet-derived endothelial cell growth factor (PDECGF) were significantly different between the recurrent and non-recurrent groups (p &ly; 0.1; Wilcoxon-Mann-Whitney analysis). Chi-squared test showed significant differences in MIB-1, cdc2 and stromal plasminogen activator receptor (suPAR). Histologically, mitotic count was also significantly different between the two groups (p < 0.005). Multivariate analysis revealed that positivity for cdc2 (p=0.01), high mitotic count (p=0.04) and negativity for CD44v9 (p=0.02)were independent prognostic factors among variables selected by univariate analysis, and that positivity for MIB-1 (p=0.03) and cdc2 (p =0.01), and negativity for CD44v9 (p =0.03) were independent prognostic factors among the immunohistochemical markers examined. CONCLUSION: Our results indicated that positivity for MIB-1 and cdc2, high mitotic count and negativity for CD44v9 could serve as independent factors for predicting the recurrence of node-negative breast cancer.