A small-scale filament winding apparatus was built so that surface temperature distributions could be determined during actual filament winding runs and the resulting temperatures could be compared to those predicted by a numerical model. Several experimental runs were performed using various combinations of infrared energy input and prepreg tape velocities. Infrared thermography was used to determine the surface temperature of the composite during each experiment. The field of view of the infrared camera included approximately 90 degrees of the mandrel in the angular direction and two-thirds of the axial length of the composite on the mandrel. The numerical and experimental results in this paper suggest that varying the position of the infrared lamp, changing the prepreg tape velocity (winding speed), or the heat flux incident on the composite have a major impact on the temperature distribution throughout the composite; these factors may therefore be used to alter the physical properties and overall quality of a filament-wound part. Improved consolidation will likely occur if a low-power infrared lamp is used to preheat the prepreg tape prior to contact with the mandrel and another infrared preheater is focused on the mandrel at an angular location prior to the tape-mandrel interface. The majority of the composite heating should be achieved by a high-power lamp located such that the energy is focused on the mandrel in the vicinity of the prepreg tape-mandrel interface but displaced in the direction opposite the axial velocity of the tape. Therefore, the location of the lamp must move relative to the tape when the crosshead feed mechanism changes directions. By displacing the axial location of the lamp slightly downstream (in the direction of motion of the crosshead feed mechanism) of the tape-mandrel interface, preheating of the composite on the mandrel occurs prior to the tape being laid on the mandrel. The numerical model is used to identify processing windows that lead to a well-consolidated part. The processing window is defined by upper temperature limits that will damage the part and lower temperature limits that will leave the part unconsolidated. The temperature limits are selected from a knowledge of the autohesive characteristics of APC-2 (complete intimate contact and void-free prepreg tape are assumed). The model is then used to identify winding conditions such as infrared energy input, winding speed, and composite substrate thickness that result in a consolidated product.