Using JTWC (Joint Typhoon Warning Center) best track analysis data for the Indian Ocean cyclones, we developed an empirical equation for prediction of maximum surface wind speed of tropical cyclones during first 6–12 hours of landfall along the coastline of Indian subcontinent. A non-linear data fitting approach, the Genetic Algorithm, has been used to develop the above empirical equation using data for 74 tropical cyclones that made landfall on the coasts of India, Bangladesh and Myanmar during the period 1978–2011. For an out of sample validation test, the mean absolute error of the prediction was found to be 5.2 kt, and a correlation of 0.97. Our analysis indicates that time-integration of land area intercepted by cyclones during the landfall is a better predictor of post-landfall intensity compared to post-landfall time span. This approach also helps to tackle the complexity of coastline geometry of Indian subcontinent area.