Pushed by increasing availability of price data and extensive market liberalisation efforts in many developing countries, research on food market integration has evolved rapidly over the last two decades. Empirical methods to measure market integration diverged in two directions: on the one hand, there is the Parity Bounds Model (PBM), while on the other hand the use of Threshold Autoregressive (TAR) Models has been proposed. This article provides a discussion of the two methods and argues that TAR models are more able to capture the dynamics of the arbitrage process underlying interconnected markets. Furthermore, we extend the standard TAR model to include a time trend in both the threshold and the adjustment parameter. Using weekly maize price data on seven selected markets in Tanzania, we illustrate how both transaction cost and the speed of adjustment have changed during the nineties.