The mouse mammary tumor virus acquires a highly reproducible chromatin structure when integrated into cellular DNA. Previous studies have suggested that the LTR is arranged as a series of six phased nucleosomes, that occupy specific positions on the LTR. On the basis of nucleosome reconstitution studies using DNA from the B region of the LTR, it has been argued that this sequence directs a uniquely positioned nucleosome. Here we demonstrate in vitro that reconstituted B region nucleosomes adopt at least five distinct translational positions in two rotational frames on a 206 bp fragment of DNA. We have resolved an initial reconstitute into its component species using nondenaturing gel electrophoresis, and precisely mapped the positions of each species using a hydroxyl radical footprinting assay. To confirm the nucleosome positions determined with the hydroxyl radical assay, nucleosome boundaries were mapped using exonuclease III. Comparison of the results from the hydroxyl radical footprinting and exonuclease III assays revealed a symmetrical pattern of overdigestion by exonuclease III which made unequivocal determination of nucleosome boundaries dubious. We conclude that the general use of exonuclease III to map the positions of nucleosomes may lead to incorrect assignment of position, and that assignment of position through the determination of the nucleosome pseudo-dyad from hydroxyl radical footprinting data represents a superior method of analysis.