Abstract: This study analyzes the overall size distribution of cities and towns of India in 2011. The size distribution of urban agglomerations does not follow the Zipf’s law. It is shown that lognormal (LN) distribution is fair at best and there are notable deviations, which occur over the entire range of cities and towns. Then it have attempted to fit Double Pareto Lognormal (DPLN) distribution for the entire range of cities and towns. The fit clearly indicates that DPLN fits the data much better than LN based on the computed values of Akaike Information Criterion (AIC) and Bayes Information Criterion (BIC). It has also compared the model performance using LR tests and Bayes factors. The results are consistent with the earlier findings for city sizes in the US and other countries.