By equilibrium competition experiments, the dissociation constant (K(RD)) of lac repressor for E. coli DNA carrying a deletion of the lac operon was measured at a variety of salt concentrations. These data are used in the consideration of several aspects of protein-DNA interaction: Quantitative estimates of specificity are made. Specificity changes only slightly with salt concentration. We calculate that in vivo, 98 percent or more of repressor is bound to DNA predominately at sites other than the lac operator. Inducers shift repressor from operator to nonoperator DNA, but do not free it from DNA. The general affinity of repressor for E. coli DNA is sufficient to support a model where repressor slides along DNA for significant distances. The effective dissociation constant of repressor for operator (K(eff)) is very sensitive to the total DNA concentration. We propose that "junk" DNA in eucaryotes functions to maintain total DNA at an optimum concentration. We consider the lac operon in the nucleus of a lymphocyte, point out that severe difficulties would be encountered, and suggest possible solutions.