Based on the cyclic statistics

a non-dimensional index of degree of cyclostationarity （DCS） for machine condition classification is proposed. The index was designed to be a ratio of the cyclostationary and the stationary component of a signal. Since the statistics of some mechanical signals may be cyclostionarity

DCS is then developed to measure the cyclostationarity. Applied the index to classify the conditions of rolling bearing

the effectiveness was proven. Furthermore

compared with the traditional dimensionless indexes

such as impulse factor and kurtosis value

the effect of DCS is much better

especially for the condition of ball flaking. Thus DCS provides a novel tool to diagnose machine.