Urban poverty is complex and conventional money-metric poverty fails to measure the multiple deprivations of the urban population. Though recent estimates of multidimensional poverty do capture multiple deprivations, they do not capture the extent of multidimensional poverty in urban India. Using the urban sample from the National Family Health Survey, 2015-16, this paper estimates and decomposes multidimensional poverty in urban India. Urban poverty is measured in four key domains: Education, health, standard of living, and housing. A multilevel logistic model is used to decompose the variations in multidimensional poverty across geographical regions. Results suggest that about one-third of the urban Indian population is multidimensionally poor and one-sixth is vulnerable to multidimensional poverty. The state patterns of multidimensional poverty were diverse, with more than half of the urban population in Manipur and Bihar being multidimensionally poor, followed by Nagaland and Uttar Pradesh. On controlling for household characteristics, 17.5% of the total variation in multidimensional poverty was attributable to census enumeration blocks, 6.6% to districts, 1.8% to regions, and 9.9% to states. The odds of multidimensional poverty were higher among large households, female-headed households, widowed, and scheduled tribes. Contextualizing multidimensional poverty and prioritizing vulnerable groups and regions are essential for reducing multidimensional poverty in urban India.