This paper presents a new method to identify unknown parameters of linear fractional order systems by discretizing it at the Gauss-Lobatto-Chebyshev collocation points. The proposed spectral parametric estimation method benefits from the spectral method exponential convergence feature and results in more accuracy in the estimated parameters and less computational time. The advantages of using the spectral parameter estimation method are shown in two examples. In the first example, a batch of five isothermal creep experimental data for an epoxy is used to estimate a simple solid viscoelastic model with fractional order. In the second example, the state matrix and fractional orders of a linear system with non-commensurate fractional order are estimated by using its input-output data disturbed by unknown white Gaussian measurement noise.