Activities of the first three enzymes in the de novo purine biosynthetic pathway have been measured in cell-free extracts of the Chinese hamster ovary cell (CHO-K1) and two purine-requiring auxotrophs of this cell. Ade-A has been found to be defective in phosphoribosylpryophosphate (PRPP) amidotransferase while Ade-C has been found to be defective in glycinamide ribonucleotide (GAR) synthetase. Neither enzyme deficiency is due to the presence of an excess of diffusible inhibitor, and mixed extracts of Ade-A and Ade-C are capable of performing both enzymatic steps in a coupled assay. Assays of GAR formyltransferase show that it is present in Ade-A and Ade-C, indicating that these cell types are defective in only one enzyme each of the early purine biosynthetic enzymes. Using the Ade-A mutant, analysis of alternatives to PRPP plus glutamine as substrates for the first step in the purine biosynthetic pathway showed that a common genetic unit must direct the synthesis for both PRPP plus glutamine and PRPP plus ammonia activities. Although ribose-5-phosphate plus ammonia can be used in cell-free extracts to perform the first step in purine biosynthesis, it is shown that this activity is apparently not used by intact CHO-K1 cells.