Radiation doses necessary to control 50% of spheroids (SCD50) were determined for five human soft tissue tumor lines after single dose and fractionated irradiation. Spheroids with 1000-1500 cells were used throughout. A similar number of cells per spheroid resulted in different sized spheroids for the respective cell lines. The parameters alpha, beta, and the number of regenerating cellular units per spheroid (SRU) were estimated from the spheroid control data using a direct fit according to the linear quadratic model assuming Poisson statistics. The number of spheroid regenerating cellular units was also determined from the growth delay at doses required for 10% spheroid control. In addition, alpha, beta, and the fraction of clonogenic cells of the five cell lines were obtained from a soft agar colony forming assay. The most precise parameter for radiation sensitivity was the SCD50, with a coefficient of variation smaller than 5%. SCD50 values ranged from 5.9 to 11.0 Gy for the five soft tissue tumor lines. Two of the five cell lines showed significantly higher alpha values and lower calculated survival fractions after 2 Gy (SF2) in the soft agar clonogenic assay than in the spheroid control assay. This points to a resistance-enhancing effect in the spheroid system. Whereas the fractions of SRU from the number of cells per spheroid, estimated from the spheroid control and growth delay assays, agreed well, no significant correlation existed between the fraction of SRU and the fraction of clonogenic cells in the soft agar colony forming assay. The alpha/beta ratios as a descriptive measure of the fractionation sensitivity of the tumor cell spheroids in the spheroid control assay corresponded well with those derived from the dose-cell survival data using a soft agar colony forming assay. Two of the five cell lines showed high fractionation sensitivities with alpha/beta values smaller than 5 Gy while those of the remaining three ranged from 7.8 to 10.8 Gy. Spheroids are structurally more similar to in vivo tumors than monolayer cultures. From the observed lack of correlation in the radiosensitivity parameters alpha and SF2 as well as in the fraction of SRU or clonogenic cells obtained from the spheroid control assay or the colony forming assay, one would expect even greater differences between results from colony forming assays and the radiosensitivity of in vivo tumors, at least for human soft tissue sarcomas.