Implantable intrathecal drug delivery systems (IDDSs) have been used to manage refractory cancer pain, but there are no randomized clinical trial (RCT) data comparing them with comprehensive medical management (CMM). We enrolled 202 patients on an RCT of CMM versus IDDS plus CMM. Entry criteria included unrelieved pain (visual analog scale [VAS] pain scores >/= 5 on a 0 to 10 scale). Clinical success was defined as >/= 20% reduction in VAS scores, or equal scores with >/= 20% reduction in toxicity. The main outcome measure was pain control combined with change of toxicity, as measured by the National Cancer Institute Common Toxicity Criteria, 4 weeks after randomization. Sixty of 71 IDDS patients (84.5%) achieved clinical success compared with 51 of 72 CMM patients (70.8%, P =.05). IDDS patients more often achieved >/= 20% reduction in both pain VAS and toxicity (57.7% [41 of 71] v 37.5% [27 of 72], P =.02). The mean CMM VAS score fell from 7.81 to 4.76 (39% reduction); for the IDDS group, the scores fell from 7.57 to 3.67 (52% reduction, P =.055). The mean CMM toxicity scores fell from 6.36 to 5.27 (17% reduction); for the IDDS group, the toxicity scores fell from 7.22 to 3.59 (50% reduction, P =.004). The IDDS group had significant reductions in fatigue and depressed level of consciousness (P <.05). IDDS patients had improved survival, with 53.9% alive at 6 months compared with 37.2% of the CMM group (P =.06). IDDSs improved clinical success in pain control, reduced pain, significantly relieved common drug toxicities, and improved survival in patients with refractory cancer pain.