Recent global warming has greatly reduced the area of perennial sea ice in the Arctic Ocean, caused mountain glaciers to retreat and allowed major ice streams draining the Antarctic and Greenland ice sheets to approach and even exceed surge velocities. This has brought new urgency to the concept that large ice sheets, past, present, and future, may have the capacity to self-destruct causing rapid and permanent changes in global climate and sea level. This possibility has focused attention on these ice streams, beginning with Jakobshavn Isbrae, long the world‘s fastest ice stream and discharging seven percent of Greenland ice, a discharge rate that has now doubled following catastrophic disintegration of its buttressing ice shelf in Jakobshavn Isfjord in 1997. A geometrical solution to the longitudinal force balance was combined with a simple mass balance to obtain two solutions to the resulting force-and-mass balance from 1986 to 2006. The two solutions are obtained using two approaches to quantify the loss of buttressing. The best match to changes in surface elevations and velocities along the ice stream, especially the rapidly lowering surface during this timespan, is provided by the solution that has the strongest reduction of ice-bed coupling up the ice stream, quantified by a rapid increase in the ―floating fraction‖ of ice. This pattern seems to be repeating for other Greenland and Antarctic ice streams. If rapid surface lowering and ice discharge spreads to all major ice streams and lasts for centuries, even accelerating over time, selfdestruction of these two remaining ice sheets seems inevitable. Without the high albedo of ice sheets, global warming would increase until a new global equilibrium with the new boundary conditions was established. The other possibility is the observed rapid ice thinning and discharge may be transient and lead to a new stable equilibrium that keeps these ice sheets largely intact with no significant change in global climate and sea level. Our results from Jakobshavn Isbrae were unable to determine which scenario is most likely, but time will tell, probably over the next few decades as observational data accumulate and our modeling capability improves.