The relationship between Atlantic meridional overturning circulation (AMOC) variability and high-latitude North Atlantic buoyancy changes is complicated by the latter both driving, and responding to, AMOC changes. A maximum covariance analysis applied to a 1,201-year preindustrial control simulation reveals two leading modes that separate these two distinct roles of North Atlantic temperature and salinity as related to AMOC variability. A linear combination of the two modes accounts for most of the variation of a widely used AMOC index. The same analysis applied to another control simulation known to possess two distinct regimes of AMOC variability—oscillatory and red-noise—suggests that the North Atlantic buoyancy-forced AMOC variability is present in both regimes but is weaker in the latter, and moreover there is pronounced multidecadal/centennial AMOC behavior in the latter regime that is unrelated to North Atlantic buoyancy forcing.