The structural proteins of Sindbis virus are translated as a polyprotein precursor that is cleaved upon translation. The capsid protein is postulated to be a serine protease that releases itself from the N terminus of the nascent polyprotein by autoproteolysis. We have tested the importance in autoproteolysis of His-141, Asp-147, and Ser-215, previously postulated to form the catalytic triad of the protease, and of Asp-163. Several site-specific mutations were constructed at each of these positions, and the release of the capsid protein during translation in a cell-free system was examined. Because proteolysis occurs in cis during translation, the kinetics of release cannot be determined in this system, but the extent of proteolysis can be ascertained. Ser-215 appears to be the catalytic serine of the proteinase. Cys or Thr could substitute inefficiently for Ser-215, but substitution with Ala or Ile led to complete loss of activity. His-141 was also important for proteolysis. Substitution with Ala or Pro led to total loss of activity. Surprisingly, substitution with Arg resulted in complete proteolysis in vitro. Changes at the two Asp residues resulted in complete proteolysis of the substrate in vitro. All mutations that resulted in at least partial cleavage in vitro were incorporated into a full-length clone of Sindbis virus and an attempt was made to recover mutant virus. All changes tested were lethal for the virus except Asp-163 to Asn. Thus, production of infectious virus is either a more sensitive measure of the catalytic rate than the extent of in vitro cleavage, or these residues have necessary functions in addition to their possible role in proteolysis.