Clostridial phosphotransacetylase catalyzes acyl group transfer between coenzyme A (CoA) and inorganic phosphate and also the arsenolysis of acetyl-coenzyme A (AcCoA) to yield acetate and CoA-SH. The enzyme mobility on sodium dodecyl sulfate electrophoresis corresponds to a molecular weight of 70 000. Kinetics of both forward and reverse reactions are of the ternary type as previously reported and product inhibition data are consistent with a random binding scheme. One essential sulfhydryl group per 70 000 daltons was inactivated in a pseudo-first-order process by either N-ethylmaleimide or 5,5'-dithiobis (nitrobenzoic acid). Reduction of the rate of this inactivation by 50% in the presence of AcCoA or acetyl phosphate concentrations near their kinetic K values demonstrates binding of these acyl donors in simple enzyme-substrate complexes. Moreover, pulse-chase experiments show these binary complexes to be functional and also show that they do not dissociate rapidly compared with their rates of catalytic turnover. Incubation of the enzyme with 14C-labeled acyl donors failed to produce labeled protein after passage through Sephadex. This was true despite efforts to mimic "substrate synergism" with desulfo-CoA or to compensate for unfavorable equilibria by means of CoA traps. Very slow isotope exchange reactions of 32Pi into acetyl phosphate and [3H]CoA into AcCoA were at first observed. As in the cases of several other enzymes recently reexamined, these were shown on careful inspection to be artifacts of contamination by second substrates. Attempts to detect exchange reactions between acetyl phosphate and Pi, even in the presence of the CoA analogue, desulfo-CoA, were also unsuccessful. Therefore, no evidence for an acyl-enzyme could be detected. Furthermore, our data allow us to develop arguments which, we believe, indicate that an acyl-enzyme intermediate is extremely improbable in the reaction catalyzed by phosphotransacetylase.