Statistical and numerical modeling tools were used to investigate the climatic effects of urbanization in Istanbul, the largest city in Turkey. Mann–Kendall trend test was applied to minimum temperature data from stations located in urban, suburban and rural areas in Istanbul to determine the existence and significance of trends, and the approximate years in which changes in the trends started. In addition, using a mesoscale atmospheric model, a sensitivity experiment was carried out to explore the atmospheric effects of urbanization in the city. Both statistical and modeling analyses indicated significant warming in the atmosphere over the urbanized areas. Mann–Kendall tests indicated statistically significant positive trends in the time series of the differences in minimum temperatures between urban and rural stations. Seasonal analyses showed that the urbanization effect on climate was most pronounced in summer. In most cases, the changes in the trends occurred in the 1970s and 1980s when the population growth rate in Istanbul increased dramatically. The model results exhibited a significant expansion of the urban heat island in Istanbul from 1951 to 2004, fairly consistent with the expansion of the city in this period. A two-cell structure for the urban heat island emerged at the reference level from the difference of the July simulations with current and past landscapes: one on the European side and the other on the Asian side of the city. The maximum reference-level temperature difference between the past and present simulations was found to be around 1 °C. The modeling experiment also indicated that the velocity of the prevailing northeasterly wind and the water vapor mixing ratio were both reduced over the city. The heating effect due to urbanization was found to penetrate about 600–800 m height in the atmosphere over the city, and the two surface heat island cells were found to combine aloft. Copyright © 2006 Royal Meteorological Society