A model has been used to simulate potassium (K) uptake by wheat in a pot culture experiment. Three soils from India, namely Alfisol, Inceptisol and Vertisol, were differentially K exhausted by Sudan grass (Surghum vulgare var. Sudanensis) for a period of 280 days and were used to simulate potassium uptake by wheat (Triticum aestivum) and also to predict the amounts of K released or fixed during cropping. Except in Alfisol all the predicted values of K uptake closely agreed with that of observed values. When predicted K uptake values were plotted against the observed values, r2 values were found to be 0.927, 0.828 and 0.721 in Inceptisol, Alfisol, and Vertisol, respectively. There is a close relationship between observed and predicted values of K uptake as evident from the high r2 values, but in case of Alfisol and Vertisol the model has over-predicted K uptake, which perhaps was due to over prediction of K release from non-exchangeable K. The model has been validated and has been applied to simulate response towards fertilizer application at different available K. It has been shown that maximum response occurs at a particular value of available K which shifts towards higher value as Release Threshold Level (RTL) increases. Predicted K uptake was most sensitive to changes in root parameters such as Root Length Density (RLD) and maximum influx rate (Vmax), since changes in the time at which maximum root length density was attained (RLD-B) and the decay constant of Vmax (Vmax-B) gave the greatest changes in K uptake in almost all the soils except highly exhausted Alfisol in which predicted potassium uptake was more sensitive to changes in RLD-B and intercept (c) than to root uptake kinetics, as described by Michaelis-Menten constant (Km) and Vmax. The predicted potassium uptake was least sensitive to changes in rate constants of release and fixation and fixation threshold level in Inceptisol and Vertisol, but was sensitive to release threshold level and rate constant of release in Alfisol.