Poverty among older people in urban areas is growing along with population ageing and urbanization in China, but the absence of proper indices and method to measure urban elderly poverty brings difficulty to the understanding of the spatial pattern of elderly poverty in cities and policy making in poverty reduction. It is necessary to consider the association of elderly poverty with education, health, family relationship and social support levels of older people and set poverty cutoff more appropriately. With a factor analysis method, we develop a resource-based poverty index to identify poor older residents on the basis of survey data in eight typical residential communities in Beijing. Then, the elderly poverty rates for different residential communities are estimated and the spatial patterns of urban elderly poverty are investigated. It is found that, (1) resource, health, independence and age factors interpret the variation of older residents, among which the score of resource factor constitutes an appropriate indicator of urban elderly poverty. (2) Elderly poverty rate differs significantly for different types of residential communities. With the spatial distribution data of various communities being considered, the poverty rate of elderly population in Beijing is estimated to be 9.55%. (3) The elderly poverty rate of town centers away from the central city is the highest and the poverty rate drops down from the outer to inner areas of the central city gradually. The density of the elderly in poverty, on the other hand, goes to the highest in town centers and the inner city areas, followed by outer central city and the fringe of the central city. These results provide evidences for the formulation of urban policies against elderly poverty in the city.