One hundred fifteen patients with postsurgical stage T1-2N0 breast cancer with the minimum follow-up of 22 years and from a defined population were studied to find reliable prognostic factors predicting long-term survival rate. The 30-year survival rate corrected for intercurrent deaths was 75%, and no deaths of breast cancer occurred after the nineteenth year of follow-up. The 30-year crude survival rate was only 28%, indicating that the great majority of these patients ultimately die of other causes than breast cancer. The nuclear DNA content could be determined by flow cytometry in 95 of the 115 cases, and 52 (55%) of them were nondiploid. DNA ploidy (diploid vs. nondiploid, P = 0.11) did not have a significant influence on long-term survival rate, but cancers with the DNA index less than 1.2 had more favorable prognosis in a univariate analysis than those with the DNA index more than 1.2 (P = 0.02). The most important independent prognostic factor in Cox's multivariate analysis was the presence of lymphatic vessel invasion of cancer cells (P less than 0.001). Several other factors, including histologic grade and type, extent of necrosis, type of tumor margin, age at diagnosis, and primary tumor size could be shown to be of prognostic value in univariate and/or multivariate analyses.