Despite its importance for urban planning, landscape management, and water management, there are very few in situ estimates of urban-forest transpiration. Because urban forests contain an unusual and diverse mix of species from many regions worldwide, we hypothesized that species composition would be a more important driver of spatial variability in urban-forest transpiration than meteorological variables in the Los Angeles (California, USA) region. We used constant-heat sap-flow sensors to monitor urban tree water use for 15 species at six locations throughout the Los Angeles metropolitan area. For many of these species no previous data on sap flux, water use, or water relations were available in the literature. To scale sap-flux measurements to whole trees we conducted a literature survey of radial trends in sap flux across multiple species and found consistent relationships for angiosperms vs. gymnosperms. We applied this relationship to our measurements and estimated whole-tree and plot-level transpiration at our sites. The results supported very large species differences in transpiration, with estimates ranging from 3.2 +/- 2.3 kg x tree(-1) x d(-1) in unirrigated Pinus canariensis (Canary Island pine) to 176.9 +/- 75.2 kg x tree(-1) x d(-1) in Platanus hybrida (London planetree) in the month of August. Other species with high daily transpiration rates included Ficus microcarpa (laurel fig), Gleditsia triacanthos (honeylocust), and Platanus racemosa (California sycamore). Despite irrigation and relatively large tree size, Brachychiton populneas (kurrajong), B. discolor (lacebark), Sequoia sempervirens (redwood), and Eucalyptus grandis (grand Eucalyptus) showed relatively low rates of transpiration, with values < 45 kg x tree(-1) x d(-1). When scaled to the plot level, transpiration rates were as high as 2 mm/d for sites that contained both species with high transpiration rates and high densities of planted trees. Because plot-level transpiration is highly dependent on tree density, we modeled transpiration as a function of both species and density to evaluate a likely range of values in irrigated urban forests. The results show that urban forests in irrigated, semi-arid regions can constitute a significant use of water, but water use can be mitigated by appropriate selection of site, management method, and species.