Aspartate transcarbamylase (ATCase) of bakers' yeast has been purified 78-fold from a crude extract of a derepressed diploid strain; its specific activity was more than 300-fold that of a wild-type crude extract. During the last steps of the purification there was a parallel co-purification of carbamylphosphate synthetase (CPSase), and both activities retained full sensitivity to feedback inhibition by UTP; indeed the sensitivity of the ATCase to UTP increased during the purification doubtless due to discard of a feedback-insensitive ATCase subunit. The two enzyme activities co-eluted from gel filtration on Sepharose 6B together with the feedback site. Analytical ultracentrifugation revealed that the material was not homogeneous, showing two major peaks. Sucrose density gradient centrifugation in the presence of UTP, glutamine, and Mg2+ resulted in co-sedimentation of the two activities and the regulatory site, corresponding to a molecular weight of approximately 800 000 daltons. Omission of UTP from the gradient resulted in disappearance of the heavy peak and appearance of a new one, corresponding to a molecular weight of 380 000 and possessing both activities; the CPSase was still highly sensitive to UTP unlike the ATCase which was only slightly sensitive to retroinhibition. Omission of glutamine and Mg2+ from the sucrose density gradient caused a distinct CPSase peak to trail behind the ATCase; again, the CPSase (molecular weight 250 000) retained full sensitivity to feedback inhibition. This, together with genetic data, supports the view that the ura-2 gene which controls ATCase, CPSase, and the regulatory site is a polycistronic operon, coding for the production of two or three polypeptide chains; the CPSase subunit is inactive unless a regulatory site is present, whereas the ATCase subunit (molecular weight 140 000) is highly active but completely insensitive to feedback inhibition.