We have analyzed DNA polymerase function in repair synthesis in human diploid fibroblasts. Previous studies from many laboratories have resulted in confusing and apparently contradictory findings; our studies suggest a unifying hypothesis and indicate that previous results may not actually be in conflict. Nongrowing (confluent) cells were damaged with N-methyl-N-nitrosourea (MNU), bleomycin, X-ray, UV radiation, or N-acetoxy-2-acetylaminofluorene (NA-AAF) over a wide range of doses, and repair synthesis was studied in the presence of one of three polymerase inhibitor, aphidicolin, dideoxythymidine triphosphate (ddTTP), or N-ethyl maleimide. We find that both polymerase alpha and a non-alpha polymerase, probably polymerase beta, are involved in repair synthesis. Furthermore, there is a dose dependence for polymerase function: at low doses of damage repair synthesis is mediated primarily by the non-alpha polymerase; as the amount of damage is increased, polymerase alpha participates to an increasing extent and at high levels of damage is responsible for approximately 50-80% of the repair synthesis. In a study in rapidly growing cells, repair synthesis following a high dose of UV was refractory to aphidicolin; these findings indicate that in growing and quiescent cells DNA polymerases may function differently in repair synthesis. All of the above results may be analyzed in terms of a simple model in which two repair synthesis systems, each involving one of the polymerases, compete to put repair patches into damaged DNA.