One of the major problems in the study of DNA replications involves the presence of numerous short, non-nascent DNA chains in the cell. Such chains often contaminate nascent DNA preparations, making accurate analysis of the replicative DNA difficult. This complication can be avoided by the use of hydroxyapatite chromatography. Previous results from this laboratory had shown that short nascent DNA chains could be eluted from hydroxyapatite at low phosphate concentrations because of their non-covalent association with protein. Additional data now indicate that this isolation procedure yields a DNA preparation which contains only nascent DNA and is essentially free of non-nascent chains. 5'-Terminal labelling patterns and molecular weight distributions show that only nascent DNA chains from the DNA-protein complex and can therefore be separated from non-nascent chains if the isolation is performed under non-denaturing conditions. The protein involved in the complex appears to be rather specific for DNA chains in the replication fork since short chains resulting from the excision of dUMP immediately after replication also form the DNA-protein complex. Using this isolation technique, the 5'-terminal nucleotides of nascent DNA chains were determined in an effort to learn something about the chain initiation process. DNA was pulse-labeled in several ways, purified by hydroxyapatite chromatography, and labeled at the 5' end with [32P]phosphate. Two-dimensional thin-layer chromatography of the 32P-labeled nucleotides, obtained after enzymatic hydrolysis, revealed that all four deoxyribonucleotides are present at the 5' end in approximately equal amounts. In most cases, this 5'-terminal nucleotide distribution does not vary significantly, even when the conditions of pulse-labeling are changed. Only if the cells are allowed to run out of thymine is a variation in the relative amounts of 5'-end nucleotides detected. In this case, the amount of dTMP at the 5' end decreases significantly, reflecting the low thymine levels and indicating that the cell can compensate for this deficiency by utilizing other available nucleotides. From these results, it appears that DNA chain initiation is essentially random with respect to the nucleotide used in the initiation process.