In experiments designed to measure radiation-induced DNA damage using the DNA unwinding-hydroxyapatite chromatography technique, we observed that under some experimental conditions a significant proportion of the test DNA became tightly bound to the hydroxyapatite (HA) and could not be released even with a high concentration of phosphate buffer. Approximately 5-10% of DNA from unirradiated cells binds to the HA. With increasing radiation doses in air, the fraction of bound DNA increases, reaching about 30% at about 35 Gy. The binding exhibits many of the characteristics of a radiation-induced cell lesion: the proportion of DNA retained by the HA is less when cells are irradiated under hypoxic conditions or in the presence of the thiol radioprotector dithiothreitol; and the binding decreases when an incubation period is allowed between irradiation and harvest of the cells for assay. Studies to determine the nature of the lesion responsible for the binding demonstrated that lesion production requires a component found in cells since no binding was observed with irradiated isolated DNA or nuclear matrix; the binding is not a result of the production of DNA-protein crosslinks; and the bound DNA is single-stranded, based on its sensitivity to nuclease S1. Because of the dose dependence of the binding of DNA to HA, the slopes of the dose-response curves for DNA damage determined with this assay depend on the method used to calculate the fraction of double-stranded DNA. Our demonstration that the bound DNA is single-stranded guides the choice of the method for data analysis.