The strength of the Atlantic meridional overturning circulation (AMOC) is believed to be associated with changes in surface buoyancy in the subpolar North Atlantic, which naturally leads to a notion that the AMOC has been weakening under global warming. Yet, a variety source of observations and its assimilation into ocean circulation models have not supported such an AMOC decline so far. In this study, an aspect that has not been paid attention, regarding the maintenance of the AMOC strength, is explored: storm activity in the subpolar North Atlantic (NA). An analysis using reanalysis data shows that the wintertime turbulent heat flux variability in the LS deep convection region is largely controlled by a small number of extreme heat flux event days, suggesting a pivotal role of winter storms in prompting LS deep-water formation. A set of forced ocean-ice model simulations, in which synoptic winter storm activity associated with these event days is either suppressed or doubled over the subpolar NA, confirms the above analysis as the altered storm activity results in a substantial change in LS convection and the AMOC strength. These experiments also show an upward AMOC trend during the late twentieth century, the degree of which is to some extent related to the intensity of storm activity in the LS. The upward AMOC trend found in the first part of the dissertation opposes to a downward AMOC trend in the twentieth century coupled model simulations employing the identical ocean component. An analysis suggests that contrast to the ocean-ice model, storm activity in the LS convection region and associated heat flux decreases during the late twentieth century. Although there is also a buoyancy increase over the LS, the wintertime heat flux decrease appears to be a more dominant factor for a decrease in convection in the LS, as an increasing freshwater input from Arctic/Subarctic Ocean bypasses the interior LS along the western boundary current. Therefore, the downward AMOC trend in the coupled model can be linked ultimately to the decreasing storm activity over the LS. This study therefore suggests that storm activity over the major convection regions needs to be paid further attention in assessing AMOC variations, including long-term trend in response to a warming scenario, in future studies.