A hear-stable protein, which is a specific inhibitor of protein phosphatase-III, was purified 700-fold from skeletal muscle by a procedure that involved heat-treatment at 95 degrees C, chromatography on DEAE-cellulose and gel filtration on Sephadex G-100. The final step completely resolved the protein phosphatase inhibitor from the protein inhibitor of cyclic AMP-dependent protein kinase. The phosphorylase phosphatase, beta-phosphorylase kinase phosphatase, glycogen synthase phosphatase-1 and glycogen synthase phosphatase-2 activities of protein phosphatase-III [Antoniw, J. F., Nimmo, H. G., Yeaman, S. J. & Cohen, P.(1977) Biochem.J. 162, 423-433] were inhibited in a very similar manner by the protein phosphatase inhibitor and at least 95% inhibition was observed at high concentrations of inhibitor. The two forms of protein phosphatase-III, termed IIIA and IIIB, were equally susceptible to the protein phosphatase inhibitor. The protein phosphatase inhibitor was at least 200 times less effective in inhibiting the activity of protein phosphatase-I and protein phosphatase-II. The high degree of specificity of the inhibitor for protein phosphatase-III was used to show that 90% of the phosphorylase phosphatase and glycogen synthase phosphatase activities measured in muscle extracts are catalysed by protein phosphatase-III. Protein phosphatase-III was tightly associated with the protein-glycogen complex that can be isolated from skeletal muscle, whereas the protein phosphatase inhibitor and protein phosphatase-II were not. The results provide further evidence that the enzyme that catalyses the dephosphorylation of the alpha-subunit of phosphorylase kinase (protein phosphatase-II) and the enzyme that catalyses the dephosphorylation of the beta-subunit of phosphorylase kinase (protein phosphatase-III) are distinct. The results suggest that the protein phosphatase inhibitor may be a useful probe for differentiating different classes of protein phosphatases in mammalian cells.