This paper presents a simple and effective approach to synthesize new motions from a given corpus of motion data. First an index function, based on posture features of each motion frame, is introduced for indexing given motion data. Then based on the fact that motion coherence implies index coherence, a new motion with start frame fstart and end frame fend can be synthesized by finding a smooth path connecting fstart to fend in the multidimensional index space. Moreover, an algorithm for finding multiple smooth paths is presented and motion refinement mechanism is also provided to synthesize desired motion behavior. Experimental examples are given to demonstrate that the proposed algorithm can generate desired motion effectively and efficiently.