Propofol anesthesia is preferred during scoliosis surgery because it suppresses evoked potential spinal cord function less than other drugs and better enables the detection of spinal cord ischemia. In this study, we determined the difference between the true and predicted blood propofol levels during target-controlled infusions in children during scoliosis surgery. Arterial blood propofol measured concentrations (Cm) were compared with predicted concentrations (Cp) approximately every 30 minutes during the maintenance phase of anesthesia in 20 children. Whole blood propofol concentrations were measured using a point-of-care blood propofol analyzer. Anesthesia management was not affected by the study. The median performance error, median absolute performance error, wobble, and divergence were calculated. Children were aged 9 to 17 years and weighed 26.5 to 95 kg. The Paedfusor model was used in 16 children and the Marsh model in 4 children. In 154 blood propofol measurements, the mean difference between the Cm and Cp was 1.5 µg·mL (limits of agreement, -1.4 to 4.5 µg·mL), and the mean performance error was 44.7% (limits of agreement, -40.1% to 130.2%). The median performance error and median absolute performance error for the whole group were 39.8% (range, -20.9% to 103.3%) and 39.8% (range, 20%-103.3%), respectively. The performance errors improved with increase in duration of infusion (divergence, -2.2 [range, -1.03 to 0.13]). Cm was almost always larger than Cp except in 2 children who had consistently lower Cm than Cp (lowest Cm(s) were 1.74 and 1.96 µg·mL when the Cp was 3 µg·mL); both had the Paedfusor model and their body weights were 28 and 33 kg. Propofol target-controlled infusion models had poor performance characteristics in children undergoing scoliosis surgery. Point-of-care propofol assay may enable adjustment of the infusion to better achieve the intended blood level.