The present article is a study of the surface temperature distribution and its evolution during spring in Lake Ladoga, using NOAA-AVHRR Channel 4/Channel 5 (CH4/CH5) data, with special focus on a thermohydrodynamic phenomenon, called thermal bar. The split-window technique, where both constant and atmospheric-mass-dependent retrieval coefficients are tested, is used for calculating the atmosphere effects in the temperature derivation. The satellite-derived temperature data are compared with field data from six different years. The results show that on average, the satellite-derived temperatures exceed the in situ data by approximately 0.5-degrees-C. The relative advantages of images from day or night passes for the study of the thermal bar is also discussed. Surface temperature differences in the convectively mixed region (temperatures less than 4-degrees-C) show that day temperatures are on the order of 0.5-degrees-C higher than the night temperatures. However, the surface temperatures from day images seem to be more representative of the bulk temperature in a water column. A comparison of the thermal bar progression in Lake Ladoga during spring 1988 as obtained from satellite images and as predicted by a simple theoretical model shows good agreement. This result and results from an earlier study indicate that the model is a good prediction tool for thermal bar movement in large temperate lakes.