We offer the hypothesis that global scale millennial and glacial cycles may be initiated from the tropical Pacific. Part I used model results to illustrate how nonlinear ocean-atmosphere interactions in the tropical Pacific could generate variations in the field of sea surface temperature (SST) on both orbital and millennial timescales. The physics underlying these variations is essentially the same as that causing ENSO (El Niño - Southern Oscillation) variability in the modern climate. Here we argue that, as with ENSO but on paleoclimatic timescales, these changes in SST distribution will be accompanied by changes in the location of atmospheric convection, which will alter the global climate via atmospheric teleconnections. By analogy with ENSO, it is hypothesized that the cold phase will increase the glaciation over North America, increase low cloud cover, and reduce atmospheric water vapor. All tend to cool the earth either by increasing the planetary albedo or reducing greenhouse trapping. The warm phase has the opposite tendencies. We also critique the hypothesis that millennial changes are triggered by changes in the production of North Atlantic Deep Water.