1. A 3-year field study of the water relations of street trees in New York City was undertaken with the following objectives: (i) to ascertain whether there was empirical evidence for water deficits; (ii) if water deficits occur, to correlate these events with prevailing weather conditions; (iii) to determine whether tree water relations varied systematically with street exposure; (iv) to compare street microclimate with that of a nearby park; (v) to compare water relations of two common street tree species; (vi) to evaluate the effectiveness of a minimum irrigation programme. 2. Water deficits in trees, as evidenced by midday stomatal closure, occurred but with lower frequency than is commonly expected. Average predicted frequency during the summer was 13% over a 21-year period. When deficits occurred, they were demand driven rather than supply limited. Deficit periods were characterized by maximum air temperatures on the street >41 0C and a maximum atmospheric vapour pressure deficit of 75 kPa. Official weather station data collected in a nearby park indicated far milder conditions, with a maximum air temperature and vapour pressure deficit of 32 'C and 2*5 kPa, respectively. Over the midday period, the mixing ratio, a conservative measure of atmospheric water content, was 84 g kg-' lower on Columbus Avenue than in the park, indicating that the street microclimate was far drier. 3. The west side of the street received more solar radiation than the east and was usually slightly hotter than either the east or the park. Yet, trees on the west side of the street were not more water stressed. Though Fraxinus pennsylvanica transpired at higher rates and sustained more negative water potentials than Tilia cordata these differences were variable and could not be separated statistically. Similarly, Fraxinus pennsylvanica showed small but statistically non-significant responses to minimal weekly irrigation while Tilia cordata was unresponsive.