The radiosensitivity of skin fibroblasts derived from patients as measured in vitro by a clonogenic survival assay appears to correlate with the risk of developing severe late reactions to radiation. Unfortunately, these assays are clinically impractical as a predictive test for radiosensitivity. The purpose of this study was to assess the utility of two possible surrogate assays for radiosensitivity, pulsed-field gel electrophoresis (PFGE) and single-cell gel electrophoresis (comet assay), both of which can be used to measure DNA double-strand breaks. Twenty-three nontransformed human fibroblast cell lines exhibiting a range of radiosensitivities were studied with both of these assays. The results were correlated with measurements of radiosensitivity obtained as part of a larger study examining the correlation between cellular radiosensitivity and clinical response. [2-(14)C]Thymidine-labeled confluent cultures were irradiated at 1.0 Gy/min with doses of 0 to 150 Gy. After allowing 4 h for repair at 37 degrees C, cells were trypsinized and aliquots were used for preparing slides for the comet assay. After neutral lysis and electrophoresis, the slides were stained with ethidium bromide and 50 comet moments were measured for each dose. The remainder of the cells were formed into agarose plugs and, after neutral lysis, were subjected to PFGE. The fraction of activity released (FAR) from the well was measured by scintillation counting of appropriate segments of each gel lane. Cellular radiosensitivity was measured with a standard clonogenic assay at a low dose rate of 1.2 cGy/min, and the dose that resulted in a surviving fraction of 0.01 (D0.01) was calculated. The slope of the plot of comet moment as a function of dose for each cell line did not correlate with D0.01 (R = 0.36, P > 0.1). In contrast, the slope of the FAR as a function of dose had a weak inverse correlation with D0.01 (R = 0.43 and P = 0.05) such that the more radiosensitive cell lines exhibited a steeper dose response for FAR. Although the correlation between the slope of the dose response for FAR and D0.01 was weak, refinement of the PFGE technique may provide a potentially useful predictive assay for radiosensitivity.