In this randomized, double-blinded study in 60 ASA I or II adults with >20% body-surface area thermal burns, we investigated the feasibility of patient-controlled analgesia (PCA) with fentanyl for pain management during dressing changes and determined the optimal PCA-fentanyl demand dose. An initial loading dose of IV fentanyl 1 microg/kg was administered. Patients received on-demand analgesia with fentanyl (10, 20, 30, and 40 microg) whenever their visual analog scale (VAS) score was >2. Mean VAS scores in the 10 and 20 microg groups (7.73 +/- 1.33 and 7.20 +/- 1.21, respectively) were significantly higher than those in the 30 and 40 microg groups (4.47 +/- 0.83 and 3.90 +/- 0.63, respectively) (all P = 0.000). Demand/delivery ratios were significantly larger in the 10 and 20 microg groups (3.03 +/- 1.06 and 2.54 +/- 0.49, respectively) than those in the 30 and 40 microg groups (1.36 +/- 0.34 and 1.37 +/- 0.36, respectively) (all P = 0.000). VAS scores and demand/delivery ratios were comparable in the 30 and 40 microg groups (P = 0.260 and P = 0.977, respectively), which suggests comparable analgesic efficacy. There was no hemodynamic instability or respiratory depression. The optimal demand dose of PCA-fentanyl was 30 microg (5-min lockout interval) after an initial loading dose of IV fentanyl 1 microg/kg.