A commonly accepted fact in the biometrics related domain is that fusing multiple classifiers to make decisions general-ly leads to improved classification performance. Meanwhile, the search for an optimal fusion scheme remains extraordi-narily complex because the cardinality of the space of poss-ible combination strategies is exponentially proportional to the number of competing classifiers. This paper proposes a mixture of gated experts for the application of 3D face rec-ognition using an ensemble of 24 different scores. The mix-ture of gated experts is optimized by a Simulated Annealing (SA) based algorithm, and it automatically selects and fuses the most relevant similarity measures. Experimental results of 3D face recognition on the FRGC v2.0 database demon-strate the performance and stability of the proposed method. Moreover, as a learning-based approach, it also has a good robustness to the variations of training database.