We analyse historical (1900 – present) and recent (year 2002) data on New York city’s urban heat island (UHI) effect, to characterize changes over time and spatially within the city. The historical annual data show that UHI intensification is responsible for ∼1/3 of the total warming the city has experienced since 1900. The intensification correlates with a significant drop in windspeed over the century, likely due to an increase in the urban boundary layer as Manhattan’s extensive skyline development unfolded. For the current-day, using 2002 data, we calculate the hourly and seasonal strength of the city’s UHI for five different case study areas, including sites in Manhattan, Bronx, Queens and Brooklyn. We find substantial intra-city variation (∼2 °C) in the strength of the hourly UHI, with some locations showing daytime cool islands – i.e., temperatures lower than the average of the distant non-urban stations, while others, at the same time, show daytime heat islands. The variations are not easily explained in terms of land surface characteristics such as building stock, population, vegetation fraction or radiometric surface temperatures from remote sensing. Although it has been suggested that stations within urban parks will underestimate UHI, the Central Park station does not show a significant underestimate, except marginally during summer nights. The intra-city heat island variations in the residential areas broadly correlate with summertime electricity demand and sensitivity to temperature increases. This relationship will have practical value for energy demand management policy, as it will help prioritize areas for UHI mitigation.