Demand prediction for farm tractors is of interest to industry and government. Two different models were developed earlier. One termed a causal model, was similar to a model of a process driven by potential difference. The other was based on time series analysis. In this paper the performance of these is examined over a ten-year span, and their special merits discussed.