We develop a theory-based model of equilibrium exchange rates incorporating factors that have been found to matter empirically. The model provides insights into how variables should be measured and what are appropriate cross-country restrictions. We estimate this model using a panel of 12 industrial countries. The model fits the data relatively well, implying relatively fast adjustment to equilibrium and outperforming a random walk at longer horizons. Furthermore, we find that the rate of adjustment depends on the distance from equilibrium, suggesting that part of the explanation for slow adjustment is inaccurate measures of equilibrium.