AbstractIn this paper. the Identification of multlvariable continuous-time systems described by stochastic vector differential equations Is considered. It can be shown that although the noise term in the differential equation Is assum d a white one. after integrating the differential equation by using block-pulse functions scheme the noise term in the resultant algebraic equation b com s a correlated one. In order to get consistent estimates of system parameters. a prefi I tering technique is proposed to whi ten the correlated noise. A simulation example is used for illustration. Th singular value decomposition method is applied to the numerical computation and the simulation result Is quite satisfactory.