In two seven-year studies with gold compounds in dogs of both sexes, thrombocytopenia was observed after 45 to 72 months of dosing in three of 14 and two of 14 dogs in high-dose groups that received 2.4 to 3.6 mg/kg of auranofin per day orally or 0.5 to 2.0 mg/kg of gold sodium thiomalate intramuscularly once every three days, respectively. An immune basis for the disorder was suggested by the apparent consumptive nature of the thrombocytopenia (increased bone marrow megakaryocytes and large peripheral blood platelets), the response to corticosteroid therapy and the demonstration of increased platelet-associated immunoglobulin. The latter was demonstrated with a solid phase radioimmunoassay and by electron microscopy using a staphylococcal protein A-colloidal gold conjugate. Platelet-associated immunoglobulin decreased as the platelet counts rose, and in one dog monitored over periods of steroid-induced remissions and subsequent relapses, the amount of platelet-associated immunoglobulin G correlated inversely with the platelet count (r = 0.82). These findings suggest that the long-term administration of gold compounds in dogs is associated with a dose-dependent incidence of thrombocytopenia, which is immune-mediated and similar to that associated with parenteral chrysotherapy in man. The application of tests for platelet-associated immunoglobulin to canine patients with immune thrombocytopenia should be useful in the diagnosis of the disorder in clinical practice.