The advent of geolocated ICT technologies opens the possibility of exploring how people use the space in cities, bringing an important new tool for urban scientists and planners. Here we apply a functional network approach to determine land use patterns from mobile phone records. The data ubiquity and method versatility allow us to run a systematic comparison across Spanish cities of varied populations. Four major land use types are detected with their proportions, spatial organization and scaling showing strong resemblances in all cities. We introduce a model inspired in Schelling’s segregation able to explain and reproduce these results with simple interaction rules for land use types. Similarities across cities break down at local scale, where land use mixing is characteristic for each urban area.