Fixed-interval smoothing is a method of extracting information from inaccurate data. This monograph addresses problems for which a linear stochastic state space model is available, in which case the objective is to compute the linear least-squares estimate of the state vector in a fixed interval, using observations collected in that interval.