The unanimously accepted dependence of the static stability of the lower troposphere on the prevailing weather has, so far, not been expressed quantitatively. This work is the first systematic attempt at expressing this dependence quantitatively. Two weather classifications, an index expressing the activity of the atmospheric temperature inversion layers, and various other inversion statistics are considered for this purpose. Radiosonde data from Hemsby (UK) are used. Lamb's weather classification, which is used initially, shows clear differences in the statistics of inversions for different weather classes. The anticyclonic class is associated with the maximum activity, and the cyclonic class with the minimum activity. The weather class is also shown to affect the vertical variation of the activity of inversions. Differences in the activity of inversions due to the weather class are maximized in the layer 900-800 mb. Given that it is within this layer where the top of the atmospheric boundary layer is usually found, this result is of particular importance for air pollution modelling. An alternative weather classification is also used with the objective of associating inversion statistics not just to a particular weather type, but also to the physical reasons that are responsible for their formation. This classification shows that, in the absence of an anticyclone, the activity of inversions is rather independent of the prevailing weather pattern. However, when the anticyclonic category is also included, the differences become significant. As the results of this work express quantitatively the strong dependence of inversion statistics on the prevailing weather, they can be used for the creation of correction factors for many characteristics related to an inversion climatology (e.g. for the annual cycle of the activity of inversions). Such corrections are necessary in cases where the weather types in the study period do not reflect their frequency in the average climate.