Kornberg's success in unraveling the process of coenzyme synthesis established him as a biochemist by the early 1950s. It had also suggested to him that other large molecules such as the nucleic acids RNA and DNA would be synthesized in a similar way. He began his investigations into nucleic acid synthesis during the same years that James Watson and Francis Crick, among others, were trying to work out the likely structure of DNA. DNA had been shown to be the stuff of genetic inheritance. Its chemical composition was known, and Erwin Chargaff had noted that the amounts of adenine and cytosine bases always matched, respectively, the amounts of thymine and guanine bases in any given DNA sample. But nobody had a clue about how cells actually made DNA. Based on his experience with coenzymes, Kornberg guessed that DNA or RNA would be made in cells by an enzyme that would string together whole nucleotides rather than assemble smaller chemical pieces. If the nucleotides--a nitrogenous base (adenine, cytosine, guanine, thymine, or uracil) attached to a sugar (ribose or deoxyribose) and a phosphate group--were the basic building blocks, Kornberg needed to know how to make them. Several other researchers were working on the synthesis of adenine and guanine nucleotides, so Kornberg began with those of cytosine, thymine, and uracil. In this work, he moved to using microbes such as yeasts for his raw material, rather than animal tissue, and also employed the new techniques of radioisotope labeling and ion-exchange chromatography to track reactions and products.

In 1953, soon after embarking on this work, Kornberg left NIH, fearing that basic research would be given short shrift when NIH opened its new clinical center and additional disease-oriented institutes. (Ultimately these fears proved baseless, and he was a lifelong advocate of the work done there, claiming NIH as his scientific alma mater.) He became chair of the Department of Microbiology at Washington University School of Medicine in St. Louis. The faculty and research staff he assembled there would become an important part of his DNA work during the following decades, each contributing a different expertise to the research problems at hand.

Working with a team that included Robert Lehman, Maurice Bessman, and others, Kornberg began his investigation into nucleotide synthesis by looking at orotic acid, a likely precursor of uracil, because it is uracil with a carbon dioxide molecule added. By late 1953 he had verified that orotic acid was a uracil precursor, requiring several different enzymes for its conversion. The first enzyme produces PRPP (phosphoribosyl pyrophosphate). PRPP then combines with orotic acid to form orotic ribose P, using another enzyme. A third enzyme then splits the CO2 off the orotic ribose P, leaving uracil ribose P, also known as uridine monophosphate, which is a complete nucleotide. From there, Kornberg and his colleagues quickly found additional enzymes that could make three other nucleotides (those of cytosine, adenine, and guanine) using uridine or PRPP as starting points. In the course of this work, they also discovered that cells don't always make their nucleotides de novo from basic compounds; more frequently they use larger pieces of nucleotides salvaged from the breakdown of older nucleic acids, or from digested food. Enzymes called kinases move the missing pieces to these larger fragments to complete the nucleotide.

Now able to synthesize all five nucleotides (a colleague at Washington University had found an enzyme that made the thymine nucleotide), Kornberg felt ready to look for the enzymes that assemble nucleotides into RNA or DNA. For a short time, the research group worked on both nucleic acids, but in 1955 Severo Ochoa's lab announced their discovery of an enzyme that synthesized RNA (though it turned out to be only a RNA-like chain); Kornberg then focused all efforts on DNA synthesis. To find the crucial enzyme in broken cell extracts from E. coli bacteria, he added ATP, plus the appropriate nucleotides, tagged with radioactive isotopes to trace their incorporation into the nucleic acid chain, and then added DNA as a primer for the chain. It took many months to achieve a reliable trace of the synthesis with radioactive thymidine, so that the enzyme's activity could be traced, but this was accomplished in 1956. Next Kornberg had to isolate and purify the DNA assembling enzyme, which he named DNA polymerase, from the bacteria cell extract, separating it out from all the other proteins (including many enzymes that interfere with the synthesis) using a wide range of procedures. Within a year, Kornberg was able to synthesize DNA from a variety of sources with this polymerase. Two papers describing this work were submitted to the Journal of Biological Chemistry in October of 1957. The JBC referees, however, rejected the articles; some objected to calling the product "DNA", preferring the technically accurate but cumbersome term, "polydeoxyribonucleotide." One insisted that the product must be shown to have genetic activity to qualify as DNA (a criterion met by very few researchers at that point). Disgusted, Kornberg initially withdrew the papers, but they were published in the May 1958 issue of JBC, after a new editor assumed his post.

Several questions remained about the process and the products of DNA synthesis. First, Kornberg had to show that the synthesized DNA was a faithful copy of the template DNA. It was fairly easy to demonstrate that the synthetic and template DNA had equal amounts of adenine and thymine, and of cytosine and guanine; and that ratios of the A-T pairs to the C-G pairs were the same. But were the sequences of base pairs accurately copied? Kornberg and postdoctoral fellow John Josse devised a procedure for determining the frequency with which any one of the four nucleotides is next to any other in the template and the product, using radioactive labeling. This "nearest neighbor" procedure also confirmed that the two chains of the double helix run in opposite directions, as predicted by the Watson and Crick model. Another question concerned the unusual relationship between the DNA polymerase and the template that served as its substrate: did the template really direct the enzyme? Some biochemists were skeptical--there were no known cases of such a phenomenon. This question would be answered as Kornberg and others unraveled the incredibly complex chemistry of DNA replication during the next thirty years. Kornberg's next step was to synthesize DNA that possessed genetic activity, which would occupy him for nearly ten more years. In the meantime, his first DNA synthesis earned him the 1959 Nobel Prize in Medicine or Physiology, which he shared with Severo Ochoa.