BACKGROUND: All widely used psoas compartment block/catheter techniques have a common limitation: external landmarks do not accurately predict lumbar plexus depth, leaving practitioners to "guess" at what depth to stop advancing the placement needle when neither transverse process nor lumbar plexus is intercepted. We assessed the accuracy of ultrasound in estimating transverse process depth before needle insertion and prediction of actual needle-to-plexus intercept depth for psoas compartment nerve blocks and perineural catheter insertion. METHODS: Before needle insertion, ultrasound was used to estimate the depth of the transverse process lying directly anterior to the intercrestal line. If a transverse process was not directly anterior to the intercrestal line, then the process immediately caudad to the line was imaged. The ultrasound transducer remained in the parasagittal plane, perpendicular to the skin. After this measurement, the transducer was removed, an insulated needle connected to a nerve stimulator inserted in the parasagittal plane, and the depth of both the transverse process (if contacted) and lumbar plexus noted. A perineural catheter was subsequently inserted. RESULTS: Of 53 enrolled subjects, in 50 cases (94%), the transverse processes were identified by ultrasound at a median (interquartile; range) depth of 5.0 cm (4.5-5.5 cm; 3.5-7.5 cm). In 27 subjects (54%), a transverse process was positioned directly anterior to the intercrestal line, and in all of these subjects, the transverse process was intercepted with the block needle a median of 0.5 cm (0.0-1.0 cm; 0.0-1.0 cm) within the predicted depth. In all 50 subjects in whom the transverse processes were identified by ultrasound, the actual lumbar plexus depth measured with the needle was a median of 7.5 cm (7.0-8.5 cm; 5.0-9.5 cm), and the plexus depth was a median of 2.5 cm (2.0-3.0 cm; 0.2-4.0 cm) past the estimated transverse process depth by ultrasound. By ultrasound, the intersection of the middle and lateral thirds of the intercrestal line between the midline and a parallel line through the posterosuperior iliac spine was too lateral to permit needle-transverse process contact in 50% of the subjects. However, moving the transducer 0.75 cm toward the midline allowed for transverse process imaging in all subjects. CONCLUSIONS: For psoas compartment blocks/catheters, prepuncture ultrasound imaging accurately predicts transverse process depth to within 1 cm, and if the lumbar plexus is estimated to be within 3 cm of the transverse process, ultrasound allows prediction of maximal lumbar plexus depth to within 1 cm.