AbstractWe investigate how statistical properties of the rate of change R of the surface horizontal magnetic field in the United Kingdom differ during substorm expansion and recovery phases compared with other times. R is calculated from 1‐min magnetic field data from three INTERMAGNET observatories—Lerwick, Eskdalemuir, and Hartland and between 1996 and 2014—nearly two solar cycles. Substorm expansion and recovery phases are identified from the SuperMAG Lower index using the Substorm Onsets and Phases from Indices of the Electrojet method. The probability distribution of R is decomposed into categories of whether during substorm expansion and recovery phases, in enhanced convection intervals, or at other times. From this, we find that 54–56% of all extreme R values (defined as above the 99.97th percentile) occur during substorm expansion or recovery phases. By similarly decomposing the magnetic local time variation of the occurrence of large R values (>99th percentile), we deduce that 21–25% of large R during substorm expansion and recovery phases are attributable to the Disturbance Polar (DP)1 magnetic perturbation caused by the substorm current wedge. This corresponds to 10–14% of all large R in the entire data set. These results, together with asymptotic trends in occurrence probabilities, may indicate the two‐cell DP2 magnetic perturbation caused by magnetospheric convection as the dominant source of hazardous R > 600 nT/min that is potentially damaging to the U.K. National Grid. Thus, further research is needed to understand and model DP2, its mesoscale turbulent structure, and substorm feedbacks in order that GIC impact on the National Grid may be better understood and predicted.