The extent of DNA excision repair was determined in dermal fibroblast strains from clinically normal and xeroderma pigmentosum (XP; complementation group A) human donors after single or combined exposures to 254-nm ultraviolet light and 4-nitroquinoline 1-oxide (4NQO). The repair was monitored by incubation of the treated cultures in the presence of 1-beta-D-arabinofuranosylcytosine (araC), a potent inhibitor of long-patch excision repair, followed by quantitation of araC-accumulated DNA single-strand breaks (representing repair events) by velocity sedimentation analysis in alkaline sucrose gradients. The amount of repair in normal fibroblast strains increased as a function of UV fluence and reached a plateau at 15 J/m2; strand breaks were not detected when these same cultures were irradiated with as much as 60 J/m2 UV and incubated in the absence of araC, implying that an initial (incision) step is rate-limiting in the repair of UV damage. In normal fibroblasts (i) the incidence of araC-detectable lesions removed during fixed intervals following exposure to 4NQO (4 microM; 30 min) was approximately 2.5 times greater than that seen following irradiation with repair-saturating fluences (greater than or equal to 15 J/m2) of UV-rays; and (ii) the amount of repair in cultures treated simultaneously with 4NQO (0.5-6 microM; 30 min) and a repair-saturating fluence of UV (20 J/m2) was found to approach the sum of that arising from exposure to each separately. The XP cells (XP12BE) exhibited a deficiency in the removal of araC-detectable DNA lesions following exposure to either of the carcinogens. Since araC is known to inhibit the repair of alkali-stable 4NQO-DNA adducts (i.e., lesions assumed to be removed by the UV-like excision pathway) but not that of alkali-labile sites (i.e., DNA lesions operated on by the X-ray-like repair pathway), our results strongly imply that the multistep excision-repair pathway operative on UV photoproducts in human fibroblasts differs from that responsible for removing alkali-stable (araC-detectable) 4NQO adducts by at least one step, presumably the rate-limiting incision reaction mediated by a lesion-recognizing endonuclease.