By studying the protein C gene of 121 consecutive patients with symptomatic type I protein C deficiency, we detected 55 different candidate mutations in 90 cases. The mutations, 76% of which were missense changes, were distributed throughout the gene. More than half the missense mutations involved Cys, Phe, Pro, or Gly, amino acids known to affect the structure of the polypeptide chain by various mechanisms. Thus, 40% of protein C deficiencies may be caused by polypeptide chain instability rather than a lack of expression of the mutated allele; this may also account for phenotypic heterogeneity. Seventeen of the 55 different mutations were found in apparently unrelated families. Half the French families we studied bore one of these 17 mutations. The wide variety of mutations suggests that both sporadic cases and a founder effect contribute to the spectrum of protein C mutations in a given population. The differences in both unique and recurrent mutations in French and Dutch populations-the only large population samples so far studied-support this hypothesis.