In this paper we present a method for body motion analysis in dance using multiple Kinect sensors. The proposed method applies fusion to combine the skeletal tracking data of multiple sensors in order to solve occlusion and self-occlusion tracking problems and increase the robustness of skeletal tracking. The fused skeletal data is split into five different body parts (torso, left hand, right hand, left leg and right leg), which are then transformed to allow view invariant posture recognition. For each part, a posture vocabulary is generated by performing k-means clustering on a large set of unlabeled postures. Finally, body part postures are combined into body posture sequences and Hidden Conditional Random Fields (HCRF) classifier is used to recognize motion patterns (e.g. dance figures). For the evaluation of the proposed method, Tsamiko dancers are captured using multiple Kinect sensors and experimental results are presented to demonstrate the high recognition accuracy of the proposed method.