With a view to estimating climate change and its urban-induced bias in selected Turkish cities, we have used data from the period 1951 to 1990 recorded by 54 climate stations, four of which are corrected for their inhomogeneities. Two sets are produced; S1, including the large urban stations, and S2, consisting of rural, small urban and medium urban stations. Normalized Kendall trend test coefficients with a spatial prediction scheme, kriging, are used to construct spatial patterns of both sets together and separately. Results reveal a statistically significant cooling in mean temperatures mostly in northern regions and warming in minimum temperatures specific to large urban areas. Seasonal analysis shows that most of this cooling has been occurring in the summer and urban warming in the spring. The causes of cooling is investigated in relation to some air pollutants, SO2 and particulate matter (PM). Linear regressions performed on the time series resulted in a significant urban bias of 0.24 ¿C per 40 years on mean temperatures and 0.56 ¿C/40 years on minimum temperatures. In association with the above results, a decrease in the temperature range of 0.48 ¿C over the period owing to urban bias was found. A 0.24 ¿C urban bias magnitude of mean temperature trends is much greater than the results found on other three regions of the Earth . An overall average cooling in mean temperatures, -0.07 ¿C per decade, detected here is the same as Nasrallah and Balling's <1993> average result for the two grid points located over Turkey.¿ 1997 American Geophysical Union