The on-board diagnostic (OBD) technologies for automobiles have been well-developed; however, it could not be carried out on motorcycles directly since the operation conditions are quite different between automobiles and motorcycles. In this research, we propose a misfire detection strategy for motorcycles based on the characteristics of crankshaft rotational dynamics. At first, experiments were done on a 125cc motorcycle to investigate the variation of instantaneous crankshaft rotational speed in power stroke while the misfire events are injected at different engine operation conditions. In order to generate misfire events for the engine, a misfire generator is established for providing specific misfire rates. If a misfire takes place at higher engine speed, the instantaneous rotational speed will decline continuously during power stroke due to higher friction losses, which leads to the reduction of average crankshaft rotational speed as well. As for the misfire events under idle operation, the change of instantaneous crankshaft rotational speed becomes unapparent. The observed crankshaft rotational dynamics are utilized to form a misfire diagnostic strategy for motorcycles which is implemented in a F28335 digital signal processor. Vehicle test results show that the misfire events can be detected correctly for idle and higher engine speed operation.