We evaluated the performance of commercially available infant and pediatric cervical collars, both alone and in combination with commonly used supplemental devices (eg, Kendrick Extrication Device, half-spine board). One infant and 11 pediatric-sized collars were tested on mannequins representing an infant and a 5-year old child. Maximum forces generated by cooperative children were measured, then applied to the mannequins to reproduce head and neck flexion, extension, rotation, and lateral motion. Limitation of motion was measured in each direction for each collar and combination method. In general, collars of rigid plastic construction performed better than did foam types. However, when used alone none of the collars provided acceptable immobilization, with even the best allowing 17 degrees flexion, 19 degrees extension, 4 degrees rotation, and 6 degrees lateral motion. When combined with supplemental devices, immobilization to 3 degrees or less in any direction could be achieved. Findings were verified using cooperative children and selected collars. Overall, combination methods were more effective than cervical collars alone (P less than .001) or supplemental devices alone (P less than .05). The modified half-spine board used with a rigid collar and tape was the most effective combination method. We conclude that prehospital cervical spine stabilization in pediatric patients is best accomplished using a rigid-type cervical collar in combination with supplemental devices as described.