To study the effects of cloud cover on the magnitude and sign of the temperature bias, we used hourly observations from nine selected US meteorological stations for 1951-1999 to quantitatively evaluate the effect of cloudiness on the observed climate. A new statistical technique, recently developed to analyze seasonal and diurnal cycles in climatic trends (Vinnikov and Robock, 2002, GRL, Vinnikov et al., 2002a, GRL, Vinnikov et al., 2002b, JGR, in press) , has been applied to study the seasonal and diurnal variation of probability of clear and overcast skies, and mean surface air temperature under clear and overcast sky conditions. We find that the probability of clear sky is quite low (about 10-20%) and has been decreasing during the past 50 years at eight of the nine stations. The diurnal cycle of surface air temperature is much stronger for clear skies compared to that overcast skies. This is not surprising, but our technique allows us to precisely quantify this effect. The seasonal and diurnal relationship between cloudiness and temperature is very complex and has many interesting details. The clear-sky, summer, daytime temperature is 4-6°C warmer than that for overcast skies, but not more than 2°C cooler at night. In the winter, however, it is warmer when overcast both at night and during the day. Evidently, both radiation and advection affect this relationship. On average, the surface air temperature in the US is slightly warmer for overcast skies as compared to clear skies, with a difference of 0.7-2°C warmer at six of the stations, close to 0°C at two of them, and 0.6°C colder at one station - Seattle, WA. We will also present further analysis of the relationship of other climatic elements to cloudiness to attempt to diagnose the physical relationships involved.