OBJECTIVE: To evaluate the diagnostic accuracy of inferior petrosal sinus sampling (IPSS) in the differential diagnosis of ACTH-dependent Cushing's syndrome as compared with pituitary imaging techniques. DESIGN AND METHODS: We retrospectively studied the diagnostic accuracy of basal and post corticotropin-releasing hormone (CRH) IPSS, magnetic resonance imaging and computed tomography in distinguishing pituitary from ectopic ACTH secretion in 97 Cushing's syndrome patients: 74 with Cushing's disease (CD) and 10 with ectopic ACTH secretion (EAS). Thirteen patients were excluded because of unconfirmed diagnosis. The difference between IPSS and pituitary imaging techniques in the correctly localized pituitary adenoma in the patients with CD was also investigated. RESULTS: The basal ACTH inferior petrosal sinus:periphery (IPS:P) ratio was > or = 2 in 63/74 patients with CD (85%), and in 1/10 EAS patients (10%); after stimulation with CRH, the ratio was > or = 3 in 60/68 patients with CD (88%) and < 3 in all patients with EAS. The basal and post-CRH ACTH IPS:P ratios had a diagnostic accuracy of 86% and 90% respectively. The diagnostic accuracy of IPSS with both ratios was significantly higher than magnetic resonance imaging (50%) and computed tomography (40%). The IPS:P ratio suggested by receiver-operator characteristic (ROC) analysis that better distinguished CD from EAS was 2.10 for the basal and 2.15 for the post-CRH ratios. Using these cut-offs, the specificity of basal ratio and the sensitivity of the post-CRH test rose to 100% and 93% respectively. Diagnostic accuracy remained substantially unchanged for the basal ratio (87% vs 86%), while it rose from 90% to 94% for the post-CRH ratio. The sensitivity of IPSS was significantly higher than that of magnetic resonance and computerized tomography. IPSS was less reliable in identifying the adenoma site found at surgery than magnetic resonance imaging or computed tomography (65% vs 75% and 79% respectively). CONCLUSION: In conclusion, IPSS improved the diagnostic performance of imaging techniques. It can help in excluding transsphenoidal surgery in EAS patients. More striking results were obtained when a > or = 2.1:1 basal ratio or a > or = 2.15:1 post-CRH ratio were considered as criteria to distinguish between patients with CD and EAS. To establish correctly the location of the pituitary adenoma, IPSS is less reliable than imaging techniques.