The rejoining of gamma-ray-induced DNA double-strand breaks (DSBs) in mammalian cells was measured after various doses of gamma rays by using a version of pulsed-field gel electrophoresis to elute fragments of DNA from an agarose plug into the lane of an agarose gel. Two approaches for measuring the kinetics of DNA repair were compared. In the first method, cells are irradiated and incubated at 37 degrees C in monolayers, after which the cells are suspended in agarose and DNA is isolated and subjected to electrophoresis. In the second approach, cells are suspended in agarose first, then irradiated and incubated for repair, and the DNA is isolated for electrophoresis. In both methods the kinetics of repair appears to be biphasic, with an initial fast phase and a second slow phase. At equal doses the t1/2 for fast repair is two-fold less in cells incubated in monolayers than in cells suspended in agarose (11 min compared to 20-23 min) and threefold less after subtracting the slow repair component. In the agarose method the t1/2 values for fast repair increase with increasing radiation dose, while in the monolayer method they are constant. In both methods t1/2 values for slow repair are approximately constant with radiation dose. At a given radiation dose, the level of initial damage is two- to threefold higher as assessed by the agarose method than by the monolayer method in which DNA repair can occur during the preparation of samples. The detection of higher levels of initial damage by the agarose method permits DNA repair to be assayed at doses as low as 8 Gy and enables fast repair processes to be assayed more readily. However, in the monolayer approach, repair occurs under normal growth conditions and is not subjected to the effects of prior manipulations and/or the rates of nutrient diffusion. Thus this approach might be more representative of normal intracellular repair kinetics.