Excision repair seems to reach a plateau in normal human cells at a 254 nm dose near 20 J/m/sup 2/. We measured excision repair in normal human fibroblasts up to 80 J/m/sup 2/. The four techniques used (unscheduled DNA synthesis, photolysis of BrdUrd incorporated during repair, loss of sites sensitive to a UV endonuclease from Micrococcus luteus, and loss of pyrimidine dimers from DNA) showed little difference between the two doses. Moreover, the loss of endonuclease sites in 24h following two 20 J/m/sup 2/ doses separated by 24h was similar to the loss observed following one dose. Hence, we concluded that the observed plateau in excision repair is real and does not represent some inhibitory process at high doses but a true saturation of one of the rate limiting steps in repair.