Covalently closed intracellular and viral simian virus 40 (SV40) deoxyribonucleic acid (DNA) were separately isolated from infected African green monkey cells (BSC-1) grown in culture. The two DNA species form overlapping bands centered at different positions in a propidium di-iodide-cesium chloride (PDI-CsCl) buoyant density gradient capable of separating closed DNA species with different superhelix densities. When the dense side of a (32)P-labeled intracellular DNA band was mixed with the light side of a (3)H-labeled intracellular DNA band and again centrifuged in a PDI-CsCl density gradient, two overlapping bands formed with modes displaced from each other. Similar band-splitting experiments performed with viral DNA always gave superimposable bands. The foregoing experiments demonstrate that the intracellular DNA is heterogeneous in superhelix density, whereas, by the same criteria, the viral DNA is homogeneous. The mean superhelix density of the intracellular closed DNA is approximately three-fourths as large as the superhelix density of the viral DNA. These results rule out the possibility that closed SV40 DNA is drawn randomly from the intracellular pool and assembled without a further nicking-closing step into virions. When the cells were grown and infected in the presence of ethidium bromide (EB), the intracellular closed DNA was found to be homogeneous in superhelix density and to have the same superhelix density as the viral DNA which, in turn, was unaffected by the presence of the drug. The foregoing results were explained by postulating that the intracellular DNA is formed with a homogeneous superhelix density and becomes heterogeneous in the absence of EB as a result of a nicking-closing cycle that occurs in a spacially or temporally heterogeneous environment. The drug EB would inhibit this action by inhibiting the nicking enzyme(s).