Abstract: We present a two-tier sketch-based engineering part retrieval system enhanced with classifier combination. Given a free-hand user sketch, we propose to use an ensemble of classifiers to estimate the likelihood of the sketch belonging to each category by exploring the strengths of individual classifiers. This supports high quality part retrieval by motivating user feedback with a ranked list of top choices. Three shape descriptors have been used to generate the probability-based classifiers independently. Experiments are conducted using the Engineering Shape Benchmark database in order to evaluate the selected combination rules before we integrate the best rule for sketch classification. User studies with the system show that users can easily identify the desired groups and then the parts. In addition, the precision attained using the synthesis is better than results from independent classifiers when applied to both user sketches and 3D models.