Single injections in the anterior region of the thoracic paravertebral space (TPVS) have been reported to generate a multisegmental longitudinal spreading pattern more frequently than those in the posterior region of the TPVS. In this trial, we examined the hypothesis that a continuous thoracic paravertebral block (TPVB) administered through a catheter inserted into the anterior region of the TPVS allows a wider sensory block dispersion. Fifty consecutive patients undergoing video-assisted thoracic surgery were enrolled. Before the surgery, an infusion catheter was inserted into the TPVS through a needle placed adjacent to either the parietal pleura (group A) or internal intercostal membrane (group P) using an ultrasound-guided intercostal transverse approach according to a randomized allocation schedule. A chest radiograph was obtained postoperatively after injection of 10 mL of radiopaque dye through the catheter. Thereafter, 20 mL of 0.375% levobupivacaine was injected via the catheter, followed by commencement of continuous TPVB with 0.25% levobupivacaine at 8 mL/h. The primary outcome was the number of blocked dermatomes at 24 h after surgery. The secondary outcomes included radiopaque dye spreading patterns, the number of segments reached by the radiopaque dye, the number of blocked dermatomes at 2 h after surgery, and pain scores. The median (interquartile range [range]) number of blocked dermatomes 24 h after surgery was 3 (2.75-4 [1-6]) in group A (n = 22) and 2 (1.5-3 [0-7]) in group P (n = 25; p = 0.037). No significant differences in the other outcomes were found between the groups. In conclusion, a continuous TPVB administered using a catheter supposedly inserted into the anterior region of the TPVS allows a wider sensory block dispersion than a catheter inserted into the posterior region of the TPVS. This trial is registered with the UMIN Clinical Trials Registry (UMIN000018578).