Abstract. We employ JRA-55 (Japanese 55-year Reanalysis), a recent second-generation global reanalysis providing data of high quality in the stratosphere, to examine whether a distinguishable effect of geomagnetic activity on Northern Hemisphere stratospheric temperatures can be detected. We focus on how the statistical significance of stratospheric temperature differences may be robustly assessed during years with high and low geomagnetic activity. Two problems must be overcome. The first is the temporal autocorrelation of the data, which is addressed with a correction of the t statistics by means of the estimate of the number of independent values in the series of correlated values. The second is the problem of multiplicity due to strong spatial autocorrelations, which is addressed by means of a false discovery rate (FDR) procedure. We find that the statistical tests fail to formally reject the null hypothesis, i.e. no significant response to geomagnetic activity can be found in the seasonal-mean Northern Hemisphere stratospheric temperature record.