Abstract Sudden stratospheric warmings (SSWs) are characterized by a pronounced increase of the stratospheric polar temperature during the winter season. Different definitions have been used in the literature to diagnose the occurrence of SSWs, yielding discrepancies in the detected events. The aim of this paper is to compare the SSW climatologies obtained by different methods using reanalysis data. The occurrences of Northern Hemisphere SSWs during the extended-winter season and the 1958–2014 period have been identified for a suite of eight representative definitions and three different reanalyses. Overall, and despite the differences in the number and exact dates of occurrence of SSWs, the main climatological signatures of SSWs are not sensitive to the considered reanalysis. The mean frequency of SSWs is 6.7 events decade−1, but it ranges from 4 to 10 events, depending on the method. The seasonal cycle of events is statistically indistinguishable across definitions, with a common peak in January. However, the multidecadal variability is method dependent, with only two definitions displaying minimum frequencies in the 1990s. An analysis of the mean signatures of SSWs in the stratosphere revealed negligible differences among methods compared to the large case-to-case variability within a given definition. The stronger and more coherent tropospheric signals before and after SSWs are associated with major events, which are detected by most methods. The tropospheric signals of minor SSWs are less robust, representing the largest source of discrepancy across definitions. Therefore, to obtain robust results, future studies on stratosphere–troposphere coupling should aim to minimize the detection of minor warmings.