The Atlantic Meridional Overturning Circulation (AMOC) of the ocean is a singular feature of the general circulation thought to play a major role in maintaining the climate of the planet. There is an intense interest in developing nowcasting and projection systems for the AMOC because of i) its association with variations in meridional ocean heat transport, North Atlantic sea surface temperatures and climatic variables such as air temperature, precipitation, drought and severe weather events such as hurricanes, (ii) its potential predictability, iii) its possible role in abrupt climate change particularly in response to anthropogenic forcing. Motivated by this background, here we propose a collaborative study between NCAR, GFDL, and MIT to: 1. Characterize modeled AMOC variability and its climate impacts: past, present, and future, 2. Identify the mechanism(s) of AMOC variability in the GFDL, MIT, and NCAR coupled models, 3. Explore the extent to which the AMOC is predictable by experimenting with prototype predictability systems initialized by ocean state estimates. Our study is of particular importance because, as the community embarks on an ambitious program of study of Atlantic climate variability, a theoretical underpinning analogous to that which motivated modeling and observations of ENSO, is still lacking. It is hoped that by capitalizing on the very significant efforts in coupled global climate modeling and state estimation methodologies at NCAR, GFDL, and MIT and drawing together their complementary strengths, we will make significant progress in each of the above foci areas.