Cyclin D1 overexpression, detected by standard immunohistochemistry, was correlated with other prognostic variables and its prognostic value was evaluated in a group of 148 invasive breast cancers with long-term follow-up. Overexpression of cyclin D1 (59% of cases) was negatively correlated (chi 2 test) with histological grade (P = 0.0001), mean nuclear area (P = 0.004), mean nuclear volume (P = 0.02), and mitotic activity (P = 0.03) and positively correlated with estrogen receptor (P = 0.0001). There was a strong correlation between cyclin D1 overexpression and histological type (P = 0.0001). Positive cyclin D1 staining was seen in 11 of 13 tubular carcinomas, 3 of 3 mucinous carcinomas, 4 of 4 invasive cribriform carcinomas, and 17 of 20 lobular carcinomas. Of 102 ductal cancers, 52 were positive, and all 6 medullary carcinomas were negative. There were no significant correlations with lymph node status, tumor size, or DNA ploidy. In survival analysis, cyclin D1 overexpression did not provide significant univariate or multivariate prognostic value. In conclusion, cyclin D1 is mainly overexpressed in the well differentiated and lobular types of invasive breast cancer and is strongly associated with estrogen receptor positivity. It is negatively correlated with the proliferation marker mitoses count and with the differentiation markers nuclear area and nuclear volume. However, cyclin D1 overexpression does not seem to have prognostic value in invasive breast cancer when no adjuvant treatment is given.