Results from this study conducted in November 1990 at the Long-Range Analysis of War Project, University of Colorado, predict that the magnitude μ of a war between Iraq and the United Nations collective security coalition will be between 5 (hundred thousand combatant fatalities) and 6 - (low million). War magnitude is defined as the common logarithm of total combatant fatalities. This falsifiable forecast (“an event 5 ≤μ≤ 6 - will occur”) was derived using mathematical models for fatalities, extent, and duration, estimated on historical populations of Correlates of War Project data on interstate wars involving great powers. The forecast will be wrong if and only if 5 > μ > 6 - . However, failure is scientifically unlikely in this case, because the models are parsimonious, they show strong empirical fit, and extensive tests prove that they are largely insensitive to historical evolution - unlike Lanchester-type models.