When 125I-labeled H6 was incubated with trout testis nuclei under conditions of pH and ionic strength approximating those in vivo, the radioactivity bound nearly quantitatively to the chromatin. Under comparable conditions, most non-nuclear proteins do not bind. Binding was neither tissue- nor species-specific, and HMG-17, a mammalian homolog of H6, behaved similarly to H6. The labeled and the endogenous H6 molecules were equivalent by several criteria: 1) Both were released nearly quantitatively upon treatment of the chromatin with DNase I, whereas neither was released by digestion with micrococcal nuclease, suggesting that the labeled molecules, like those of endogenous H6, were bound primarily to the core particles in transcriptionally competent portions of the genome. 2) Salt extraction curves were similar for both the labeled and unlabeled proteins, although about 15% of the labeled molecules were bound to the chromatin more loosely than those of the endogenous H6. Taken together, these results suggest that chromatin contains specific, well defined sites to which H6 binds. Upon increasing the concentration of H6 in the incubation mixture, progressively greater numbers of low affinity, presumably nonspecific binding sites become occupied. This observation has important implications for studies in which nucleases are employed to probe chromatin structure, since they suggest that H6 molecules released from specific, high affinity sites by the action of the nuclease might rebind more loosely to other regions of the chromatin.