The false nearest neighbors (FNN) algorithm is presented as a method for determining the proper regression vector for recreating the dynamics of nonlinear systems. This algorithm which was originally developed for the analysis of chaotic time-series, is used to determine the proper regression vector for input/output system identification and inferential prediction using time-series data. The FNN algorithm is presented, and the problem of analysing noise corrupted time-series is discussed. The choice of adjustable parameters in the algorithm and the data requirements for the algorithm are also discussed. The application of the algorithm to the identification of an electrical leg stimulation experiment and an industrial pulp digester model is presented and the results are analysed.