In the intelligent design of motorcycle, a large number of data in the simulation or physical experiment are almost not utilized to guide our design and make decision for the design, so rough set theory is introduced to the intelligent design of motorcycle. Then, aim at experimental data of the engine piston performance, rough set theory is used. An attribute reduction algorithm of decision table based on discernibility matrix and heuristic value reduction algorithm are adopted. Knowledge is extracted from the data of performance experiment of the engine piston, in order to enrich knowledge base in motorcycle intelligent design system.