The appropriate choice of imaging techniques to localize parathyroid tumors preoperatively remains controversial. We report the first prospective, blinded study to compare the efficacy of four imaging modalities in 100 patients with primary hyperparathyroidism (pHPT). Patients were examined by computer-assisted thallium 201/technetium 99m subtraction scintigraphy (TTS), computed tomography (CT), ultrasonography (US), and magnetic resonance (MR). Each study was performed and interpreted independently. Subsequent neck exploration and "curative" parathyroidectomy allowed correlation of surgical findings with imaging reports to score their accuracy. Overall sensitivities of the four imaging modalities were TTS, 73%, CT, 68%, US, 55%; and MR, 57%; with respective specificities of 94%, 92%, 95%, and 87%. Sensitivities for lesions located below the thyroid gland (thymic tongue and mediastinum) were TTS, 90%; CT, 46%; US, 44%; and MR, 50%; with respective specificities of 100%, 99%, 100%, and 94%. There was a significant increase in overall sensitivity when TTS and CT (90%, p less than 0.01) or TTS and US (85%, p less than 0.05) were used together; however, the combination of any three or even four imaging modalities did not increase sensitivity further. For small parathyroid tumors (less than or equal to 250 mg), no imaging technique had a sensitivity of more than 50%. None of the imaging studies accurately localized small hyperplastic parathyroid glands found in patients with multiple gland disease. Preoperative parathyroid imaging may not be indicated in pHPT patients undergoing first-time neck exploration because surgeons experienced in parathyroid surgery have a 93% to 96% cure rate.