Forty patients with multisystem disease and suspected systemic necrotizing vasculitis were evaluated with a protocol designed to confirm the diagnosis with sequential testing. All patients underwent initial laboratory testing. Subsequent studies were individualized to the patient starting with "safe" tests (skin, muscle, rectal biopsies) and progressing to "invasive" tests (arteriography, kidney and lung biopsies). No single laboratory study was found to have adequate predictive value. Skin biopsy, rectal biopsy, and arteriography were insensitive, nonspecific, or had poor predictive values. Muscle biopsy was the most valuable safe procedure (sensitivity, 50%; specificity, 100%; predictive value, 100%; predictive value of negative biopsy, 76%; efficiency, 64%). A diagnostic approach to the patient with possible systemic necrotizing vasculitis is described.