OBJECTIVE While inferior petrosal sinus (IPS) sampling correctly diagnoses pituitary‐dependent Cushing's syndrome if a significant ratio of plasma ACTH between the IPS and the peripheral blood is demonstrated, little has been said about the significance of a negative ratio in Cushing's disease (e.g. a false‐negative result). This study evaluates the results of IPS sampling in patients with Cushing's disease, and compares them with both imaging findings and transsphenoidal examination.DESIGN The results of IPS sampling were retrospectively compared with both imaging findings and transsphenoidal examination. IPS samples were obtained before and 2, 5 and 10 minutes after intravenous administration of 100 μg of CRH.PATIENTS Thirty‐two patients with Cushing's disease were evaluated. All subsequently underwent transsphenoidal examination of the pituitary gland.MEASUREMENTS The ratio of the ACTH concentrations at the IPS and in the peripheral blood (IPS : P ratio), and the ratio of the ACTH concentrations between the IPSs (interpetrosal ratio) were calculated. Radiographic evaluation of the pituitary gland was performed with magnetic resonance imaging (MRI, 29 cases) or computed tomography imaging (CT, 3 cases).RESULTS Transsphenoidal examination of the pituitary gland revealed a microadenoma in 27 cases. Radiological imaging showed a signal compatible with a microadenoma in 22 cases (68.8%), and correctly located the tumour at the side found at surgery in 14 of the 22 cases with positive transsphenoidal findings (MRI 13 cases, CT 1 case, overall 63.6%). Successful bilateral catheterization was accomplished in 30 patients (93.8%). Samples before and after CRH stimulation were drawn in 24 cases. No major complications were observed with the technique. IPS catheterization correctly predicted Cushing's disease (by means of a significant IPS : P ACTH ratio) in 27 of the 30 patients (90%) with basal sampling, and in 23 of the 24 cases with samples drawn before and after CRH administration (95.8%). Taking into account the 12 patients with a lateral microadenoma shown at transsphenoidal examination, IP sinus ACTH ratio was in agreement with the side recorded by the neurosurgeon in 8/12 cases (66.7%). MRi correctly located the tumour in 8/12 patients (66.7%). One patient showed no significant IPS : P ACTH ratio in any set of samples. His MRI showed no sign of a microadenoma. Two years later, another pituitary MRI evaluation showed a midline hypodense signal. The transsphenoidal examination revealed a microadenoma and the post‐operative plasma cortisol and urinary free cortisol fell to 293 nmol/l and 100 nmol/24 h, respectively.CONCLUSIONS Only when a significant IPS : P ACTH ratio is present can Cushing's disease be established by IPS sampling. The absence of a significant IPS : P ACTH ratio does not necessarily imply ectopic secretion of ACTH, nor does it exclude Cushing's disease. The results of lateralization by IPS sampling do not remove the need for a thorough transsphenoidal