The evolution of an 851-bp segment of the human immunodeficiency virus type 1 (HIV-1) genome encoding the nef open reading frame and U3/R elements of the long terminal repeat has been followed over a 4-year period in vivo and in vitro. The population of viral sequences at any given time was established by sequencing cloned polymerase chain reaction products. The samples studied were derived from the same man for whom a detailed analysis of the tat gene was previously described (A. Meyerhans, R. Cheynier, J. Albert, M. Seth, S. Kwok, J. Sninsky, L. Morfeldt-Manson, B. Asjö, and S. Wain-Hobson, Cell 58:901-910, 1989). Once again in vitro culture resulted in the selection of minor forms. Over a 4-year period in vivo, there was no obvious selection for, or outgrowth of, any particular nef or U3/R sequence. Few defective nef protein sequences were observed, which argues against nef acting as a negative regulatory factor. Although no functionally defective promoter/trans-activation-responsive elements were identified, the transactivation efficiencies varied between 0.2 and 2 times that of the control. The sequence encoding the most efficient trans-activation-responsive region did not outgrow others. The extreme genetic heterogeneity of the different samples of the locus, either in vivo or in vitro, indicates that there is no such thing as a single, distinct HIV sequence. It is suggested that different HIV-1 loci evolve independently, recombination being responsible for their uncoupling.