We used pulsed-field gel electrophoresis combined with densitometry to investigate variables influencing the measurement of DNA double-strand breaks (dsb) in human brain tumour cells and frozen tissue. Our studies showed that this system worked best when we ran the gel at 60 V for 22-25 h at 18 degrees C with a pulse time of 15 s in 1xTris-acetic acid-EDTA buffer. Because the densitometric analysis worked well only when DNA concentrations were within a certain range, we developed a quick assay using a DNA-specific dye (TO-PRO-1) to determine concentrations before making agarose plugs. When we used our optimal procedure to measure dsb in six frozen human brain tumour specimens, we found that radiation-resistant tumours had significantly more initial dsb than did radiation-sensitive tumours. The number of residual dsb could not be ascertained because the process of freezing the specimens destroyed dsb repair ability.