To systematically study the selection of radioresistant cells in clinically advanced breast cancer, a model system was generated by treating MDA-MB231 breast cancer cells with fractionated gamma radiation. A clonogenic assay of the surviving cell populations showed that 2-6 Gy per fraction resulted in a rapid selection of radioresistant populations, within three to five fractions. Irradiation with additional fractions after this initial increase did not increase the radioresistance of the surviving population significantly. Doses of 0.5 and 8 Gy per fraction were not effective in selecting radioresistant cells. To further determine the cause of the changes in radiosensitivity, 15 clones were isolated from the cell populations treated with 40 or 60 Gy with 2 or 4 Gy per fraction, respectively, and were analyzed for radiosensitivity. The average D(10) for these clones was 6.75 +/- 0.36 Gy, which was higher than that for the parental cell population (D(10) = 6.0 +/- 0.2 Gy). The operation of cell cycle checkpoints and the doubling time were similar for both the nonirradiated parental population and the isolated radioresistant subclones. In contrast, a decrease in the apoptotic potential was correlated (r = 0.7, P < 0.01) with increased survival after irradiation, suggesting that apoptosis is an important factor in determining radioresistance under our experimental conditions. We also isolated several subclones from the nonirradiated parental cell population and analyzed them to determine their radiosensitivity after fractionated irradiation. Ten fractions of 4 Gy (40 Gy in total) did not result in a significant increase in the radioresistance of these subclones compared to the irradiated cell populations. The possible mechanisms of the increased radioresistance after fractionated irradiation are discussed.