In this chapter it will be shown how to set up the Laird-Ware model (2.1) in state space form. The Kalman filter can then be used to calculate the exact likelihood, and nonlinear optimization used to obtain maximum likelihood estimates. There are two concepts needed. The first is how to concentrate the fixed coefficients, β, out of the likelihood so that only variance parameters need be estimated by nonlinear optimization. The second is how to handle the random effects.