Definition of survival and measurement of colony size in soft agar assays is important in establishing in vitro radiation survival curves. Conventionally, survival is assessed according to colony-forming ability. The distinction between small colonies that are abortive and those that are viable often involves a difficult and arbitrary choice for the investigator. We have examined the effect of different minimum colony sizes (greater than or equal to 25, greater than or equal to 50, greater than or equal to 75, and greater than or equal to 100 cells) on ionizing radiation survival curves for cells from established murine (CCL 53.1) and human (M1RW5) melanoma cell lines as well as from short-term human melanoma cell strains (C8146A, C8146C, C8161, C83-2C, C82-7A1, and C8442) and patient biopsy (83-4). Single cell suspensions were plated in the upper layer of the agar bilayer and cells were irradiated by single dose X rays. Giant cells did not form in colonies containing 50 or more cells. D0 values were highest (D0 values, from 390 to 100 cGy) for cells forming smaller colonies (greater than or equal to 25 cells, greater than or equal to 4-5 doublings) and lowest (D0 values, from 190 to 50 cGy) for cells forming larger colonies (greater than or equal to 100 cells, greater than or equal to 6-7 doublings). Therefore, apparent radiosensitivity was dependent on colony size selected for analysis. Precise measurement of colony size was important in establishing radiation survival curves because errors in determining the colony size will alter apparent radiosensitivity of cells. These results should help define the biological meaning of tumor colony growth in semisolid medium, and alter the interpretation of survival curves which measure sensitivity to agents using this assay.