As part of a diagnostic subproject of the second phase of the Atmospheric Model Intercomparison Project, covering the period 1979-1995, we are evaluating how well the current generation of general circulation models simulates relative atmospheric angular momentum (AAM), a fundamental measure of the atmosphere’s circulation that depends on the strength and distribution of the zonal winds. With up to approximately 30 models expected in this project, we will focus on the intra-ensemble spread and categorize the various models’ errors in AAM using the NCEP-NCAR reanalysis as our primary verification set. We will also compare AAM from the suite of AMIP-2 models with that from their AMIP-1 predecessors to assess model improvements. To date, output from 16 models has been archived: for these, the ensemble median value of AAM over the 17-year period is some 17% above that from reanalysis, and the spread, as measured by the models’ interquartile range, is 12% of the median value. An assessment of the seasonal and interannual signals in AAM produced by the available AMIP-2 models will be presented and summarized in diagrams such as those in Taylor (2001). AAM maxima during El Nino events were clearly simulated in the AMIP-1 models, although the relative strengths of the AMIP-1 ENSO peaks differed from those in the reanalysis. Given those results, we will focus on the verisimilitude of such interannual AAM signals in the new AMIP-2 set. Regional contributions to AAM will be assessed as well; preliminary results, for example, suggest a positive impact, particularly in the tropical midtroposphere, of new versions of some of the models, despite the excessive global AAM values of several others.