Unregulated groundwater extraction has led to declining water tables and increasing water scarcity in the Indian subcontinent. Understanding how farmers respond to this scarcity is important from multiple perspectives - equity in access, livelihoods security and resource sustainability. We present a case from the rapidly urbanizing Arkavathy sub-basin near Bangalore city in Southern India where irrigation is fully groundwater dependent. Using cross-sectional data from a stratified random sample of 333 farmers from 15 villages, we investigated the factors that determine their choice of crops under conditions of water scarcity and urbanization. Binary logit analysis showed that the high land holding farmers respond by tapping deep groundwater using borewells. Multinomial logit analysis revealed that access to groundwater, variation in the proximity to the product market (city) and labour availability influence crop choice decisions. We observe that current responses indicate what has been characterized in literature as chasing strategies. These largely favour the well-off farmers and hence inequitable. While choice of water intensive crops and unregulated pumping have aggravated water stress, the uptake of water saving technologies among irrigated farmers has been low, showing that resource sustainability may not be a concern where non-farm diversification opportunities exist. Acknowledgement : This article is an output of the socio-hydrological research project, 'Adapting to Climate Change in Urbanizing Watersheds' (ACCUWa) supported by International Development Centre (IDRC), Canada (grant number 107086-001). The authors thank Sanjeev D Kenchaigol and Kadambari Anantram for research support, Guruswamy, Kumar, Manjunatha, Nagendra and Sathish for field assistance and Shruthi for data management. Chandra Sekhar Bahinipati and participants in dissemination workshops in Bangalore and New Delhi provided useful comments. Thomas, Lele and Srinivasan designed the study, Thomas and Eswar led the field research, Patil, Lele and Thomas developed the analytical model, and Patil, Eswar and Thomas performed data analysis. Patil is currently affiliated to International Rice Research Institute-India, Bhubaneswar, India and Eswar is affiliated to Tata Institute of Social Sciences, Mumbai.