We prospectively studied the use of intercostal EMG monitoring as an indicator of the accuracy of the placement of pedicle screws in the thoracic spine. We investigated 95 thoracic pedicles in 17 patients. Before insertion of the screw, the surgeon recorded his assessment of the integrity of the pedicle track. We then stimulated the track using a K-wire pedicle probe connected to a constant current stimulator. A compound muscle action potential (CMAP) was recorded from the appropriate intercostal or abdominal muscles. Postoperative CT was performed to establish the position of the screw. The stimulus intensity required to evoke a muscle response was correlated with the position of the screw on the CT scan. There were eight unrecognised breaches of the pedicle. Using 7.0 mA as a threshold, the sensitivity of EMG was 0.50 in detecting a breached pedicle and the specificity was 0.83. Thoracic pedicle screws were accurately placed in more than 90% of patients. EMG monitoring did not significantly improve the reliability of placement of the screw