The problem of the identification of linear time invariant(LTI) distributed order systems in time-domain is firstly discussed, and a subspace identification algorithm based on principle component analysis(PCA) is proposed. The fractional order filter is designed to reconstruct the basic input and output equation. The principle component analysis is used to identify coefficient matrixes of distributed order systems. The cost function is utilized to transform the identification of fractional differential orders into multivariable optimization problem. An instrumental variable is chosen to eliminate the bad influence of random noises during the identification process. A numerical example is given to show the effectiveness of the proposed algorithm.