In this study, we propose using an endogenous cluster identification method due to Phillips and Sul to test agricultural market integration over a large number of markets, which also helps us to carry out further analysis to identify drivers of market integration. We explore agricultural market integration for paddy, wheat, chana, groundnut, onion, potato, and tomato across a large number of markets in India and find that markets are not fully integrated for any of these commodities. Prices converge into multiple but different numbers of clusters, indicating that the extent of integration is different across all commodities. Robustness checks show that the results are broadly on the same lines as time series methods. Based on the observed spatial patterns of price clusters, we explore the role of spatial dependence and locational factors on market integration. We find that location is an important driver and the likelihood of market integration is higher if the neighbouring district is also integrated, regardless of commodity characteristics. Additionally, we find that other variables like the number of markets, connectivity, transport, metalled roads and area under the crop significantly influence market integration