Ionizing radiation induces prompt single-strand breaks and double-strand breaks in DNA. In addition, labile sites are induced that can be converted to breaks by heat or mild alkali. When such labile lesions are present within multiply damaged sites, additional double-strand breaks can form. Current protocols for measurement of DNA double-strand breaks involve a lysis step at an elevated temperature, and consequently breaks from heat-labile sites will be generated during lysis and will be included in the measurement. However, such sites may not develop into breaks within the cell and therefore may not need DNA double-strand break repair processes for elimination. We present here a new lysis and pulsed-field gel electrophoresis protocol that is carried out entirely at 0-4 degrees C and thus avoids inclusion of heat-labile sites in the measurement. The new recommended lysis procedure involves two steps: The first step includes proteinase K, which has sufficient activity at 0 degrees C to support lysis, and the second step includes a high-salt buffer to further free the DNA from proteins and other cellular structures. Using various tests, we conclude that lysis is sufficient with this procedure to allow accurate determination of double-strand breaks by pulsed-field gel electrophoresis. Using the new protocol, it was found that heat-labile sites account for 30% of the initial number of double-strand breaks measured by conventional protocols after exposure to low-LET radiation. In addition, we show that heat-labile sites that can be converted to double-strand breaks are repaired with fast kinetics and are almost completely eliminated after 1 h at 37 degrees C. A study of cells deficient in nonhomologous end joining reveals that the residual fast repair response typically seen in such cells is solely due to repair at heat-labile sites and is not due to repair of prompt DSBs.