Conventional electrodiagnosis may localize an ulnar neuropathy to the general region of the elbow. Separating retroepicondylar compression from compression by the humeroulnar aponeurotic arcade from compression by the deep flexorpronator aponeurosis is more difficult. In 35 patients, we compared localization by conventional inching (stimulating stepwise around the elbow searching for focal conduction block or differential slowing) to localization by a more quantitative short segment incremental stimulation (SSIS) technique assessing latency change over consecutive 1 cm segments. Results of percutaneous studies were compared with findings of intraoperative electroneurography. We conclude that SSIS identifies compression levels more accurately than standard inching, and that SSIS and intraoperative electroneurography correlate highly, but not perfectly. Studies confined to a search for conduction block or differential slowing are limited by the low incidence of conduction block in chronic compression neuropathies.