Measurement of autoradiographic grains produced by the decay of incorporated radioisotopes is often used for a quantitative assay of the rate of DNA replication and DNA repair in cells or tissues. However, visual grain counting by microscopic observation is time-consuming and tedious process. Recently, Kraemer et al. reported that automated measurement of grains in cultured human cells may be facilitated by using appropriate grain counting instruments. Under their experimental conditions using Kodak NTB-3 emulsion, instrument-determined grain number per nucleus was proportional to visual counts up to 30 grains, and then leveled off at much larger visual counts. The saturation phenomenon was due to counting-loss by the instrument caused by overlapping of neighboring grains. To prevent the counting-loss, we have used in the present study Japanese Sakura NR-M2 emulsion which is less sensitive to radiation exposure than Kodak NTB-3, thereby yielding smaller size of grains per radioactive decay. Samples were prepared from cultured skin fibroblasts derived from normal individuals and xeroderma pigmentosum (XP) patients defective in DNA repair. These cells were irradiated with 254 nm UV incubated for 3 h with culture medium containing 3H-thymidine, and autoradiograms were made by dipping in Sakura NR-M2 emulsion. The number of grains as well as grain surface area per nucleus was measured by using ARTEK CYTO TALLY MODEL 900 counting instrument, and compared with visual counts. The results showed that, under our optimum condition, the instrument-determined number of grains was directly proportional to visual counts, at least up to 150 grains per nucleus, with a correlation coefficient of 0.971.(ABSTRACT TRUNCATED AT 250 WORDS)