Production of 10-base multiple DNA ladder fragments during DNase I digestion of chromatin is explained by a model which does not involve site-specific nicking by the DNase I. This model was tested because it explains why 10-base (actually 10.4 base) multiple-related fragments are paradoxically generated by both endonucleolytic (DNase I) and exonucleolytic (exonuclease III) mechanisms. This new model also explains the phenomenon of substantial single-stranded DNA production during DNase I digestion of chromatin. The latter phenomenon has been widely observed but is not explained by previous models. The single-stranded gap model to be presented makes testable predictions. Primarily, these are that DNase I produces single-stranded gaps in chromatin DNA and that the termini of 10-base multiple ladder fragments are separated by single-stranded gaps. Single-stranded gap production by DNase I was confirmed by a number of methods. Sensitivity of ladder band components (from DNase I but not staphylococcal nuclease digests) to S1 nuclease suggested that the ladder fragments themselves may compose a significant portion of these gaps. Separation of ladder fragment termini by single-stranded gaps was verified by demonstrating both resistance to the nick-specific NAD+-dependent ligase and sensitivity to T4 ligase which can ligate across gaps. Many single-stranded gaps, occurring both individually and clusters, were observed by electron microscopy using either cytochrome c labeling (where the gaps) are thinner than duplex) or gene 32 protein labeling (gaps thicker than duplex). Gap sizes were estimated by protecting them with gene 32 protein and digesting away unprotected duplexes. By this method, gap sizes fall into a ladder distribution (from 10 or 20 bases up to 120 bases), which, at least in the region of the shorter sizes, clearly indicates the sizes of single-stranded gaps formed in chromatin by DNase I.