Agricultural land needs to be protected for food production. Our objective is to provide a decision support for protecting the agricultural land in Lahore, Pakistan. To do so, first we classified the Land Use and Land Cover (LULC) from Landsat images for the years 2009 and 2012. Second, we performed Markov chain analysis to simulate the LULC change over time. The resultant probability of LULC inter conversion was then combined with the Cellular Automata (CA). Third, the spatio-temporal patterns of LULC change from CA-Markov were integrated with the land fitness map obtained through the analysis of soil chemical properties. We observed a gradual increase in built-up land and a decrease in agricultural land from years 2009 to 2012, with an increase of 18.8% to 60.3% in the built- up land, and a decrease of 43.5% to 35.9% in the agricultural land. The Markov-CA analysis further predicted a significant (p = 0.69) LULC change from year 2012 to year 2015, with an increase of 2% in built-up land, and a decrease of 1.3% in agricultural land. The resultant map shows zones to be predicted on priority bases, which can be useful in making comprehensive land policies to protect agricultural land and to secure food in developing countries.