Introduction. A new triggered electromyography test for detection of stimulus diffusion to intercostal muscles of the contralateral side during thoracic pedicle screw placement was evaluated. Experimental research was carried out in order to determine if, using this test, neural contact at different aspects of the spinal cord and nerve roots could be discriminated. Methods. Nine industrial pigs (60–75 kg) had 108 pedicle screws placed bilaterally in the thoracic spine (T8–T13). Neural structures were stimulated under direct vision at different anatomic locations from T9 to T12. Recording electrodes were placed over the right and left intercostal muscles. Increasing intensity of the stimulus was applied until muscle response was detected at the contralateral side (diffusion phenomenon). After this first experiment, the thoracic spine was instrumented. Screws were placed in the pedicle in two different positions, the anatomic intrapedicular location and with purposeful contact with the neural elements. Results. Response thresholds to direct stimulation of nerve root at different points were significantly lower than those obtained by stimulation of the dorsal aspect of the spinal cord (0.44±0.22 mA vs 1.38±0.71 mA). However, a 24-fold stimulation intensity (6.50±0.29 mA) was necessary to obtain diffusion of the EMG response to the opposite left side if the right nerve root was stimulated. Only a 2-fold increment (3.17±0.93 mA) was able to elicit diffusion of EMG responses to the contralateral side when stimulation was applied to the dorsal aspect of the spinal cord. Contralateral EMG responses after high increases of stimulation thresholds indicated nerve root contact. Diffusion phenomenon after low threshold increments reflected medullar contact. Electromyography recordings after triggered stimulation of the screws showed that only screws in contact with the spinal cord had significantly lower responses (2.72±1.48 mA). Conclusion. Stimulus-triggered EMG could only discriminate screws with violation of the medial pedicle wall if they were contact with neural tissues. Recording EMG-potentials at the contralateral paraspinal muscles (stimulus diffusion phenomenon) proved to be a reliable method to discriminate which of the neural structures was at risk