The comparison of a laboratory freeze–thaw test with a field exposure test has long been of interest to users, builders, designers, engineers and researchers for the prediction of time-dependent stability or performance of rocks as building materials. Several procedures and standards have been developed to simulate freeze–thaw (FT) effects on rock samples in laboratory conditions. However, all these procedures have different test conditions and they do not relate to the real outdoor performance of building rocks. In this study, a steel storage tank with three sections was designed to examine decomposition of rocks under atmospheric effects. Water drainage was allowed in one of the sections, which had holes in the bottom and sides, whereas the other two sections were semi-drained and undrained, respectively. Ankara ignimbrite and Golbasi (Ankara) andesite were exposed to atmospheric effects for 2 years and 4 years, respectively. The ignimbrite samples, especially those in the undrained section of the storage tank, had totally disintegrated by the end of the 2 year period. On the andesite core samples, some small cracks in the sample edges were observed at the end of 4 years. Ignimbrite and andesite samples collected from the three sections of the tank at 6 month periods were analysed in the laboratory for changes in physical and mechanical properties. A site-specific meteorological approach was used to design a new laboratory FT test procedure. The ASTM D5312 method and a new FT laboratory test procedure with a different test temperature range (−4 °C and +5 °C for the Ankara region) were used in the laboratory to simulate FT effects on ignimbrite and andesite NX type (diameter 54 mm, length-to-diameter ratio 2.5) core samples. A new Turkey FT cycle severity index map was developed to determine the freezing and thawing cycle for each test. After laboratory FT tests and outdoor exposure tests, the physical and mechanical properties of samples were determined. The changes in physical and mechanical characteristics of ignimbrite and andesite samples subjected to FT using the laboratory ASTM D5312 method were significantly different from those seen in a natural environment. The changes in physical and mechanical properties of the samples tested according to the new FT test procedure were similar to those of the samples in the undrained section of the storage tank. The effects of water content of andesite samples on physical and mechanical properties were analysed during the FT process. It was concluded that at least 50% water content is needed for disintegration and the first micro-crack development in samples, and that most of the sample destruction during FT occurs when the water content of the samples is ≥75%. The changes in physical properties of the samples subjected to the FT laboratory test with 75% water content were similar to those of the samples stored in the semi-drained section of the storage tank in natural weather conditions.