The Asian-Australian Monsoon (A-AM) anomalies were evaluated in ensemble runs of 11 atmospheric general circulation models for the period of September 1996-August 1998. Low-level circulation indices are calculated for the Indian, western North Pacific-East Asia (WPEA) and Australian monsoon region, respectively. The models exhibit faithful skill in reproducing the Australian and WPEA monsoon indices; yet fail the Indian monsoon index. A suite of 4 runs using Seoul National University model (different only in their initial conditions) for the period of 1950-1998 was also examined. In this 49-year ensemble runs, the skill of simulating Indian monsoon index improves significantly while that for Australian and WPEA monsoons decreases, suggesting that during the unprecedented strong 1997/98 El Nino, the models experienced extraordinary difficulty in simulation of the Indian monsoon low-level circulation anomalies yet gained remarkable capacity in simulating Australian and WPEA monsoon variability. While the map correlation coefficients between the observed and simulated rainfall patterns in the El Nino region (30S-30N, 160E-80W) range from 0.6 to 0.9, the rainfall pattern correlation coefficients in the Asian-Australian monsoon region (30S-30N, 40E-160E) vary from 0.25 to 0.4 in most models. These inept skills arises primarily form the miserable performance over the Southeast Asia and western North Pacific Ocean (5N-30N, 80E-150E) where the observed and simulated rainfall anomalies are virtually unrelated. It was further revealed that the observed monthly mean rainfall anomalies are negatively correlated with, whereas the simulated rainfall anomalies in the models are positively correlated with the local SST anomalies in the aforementioned oceanic regions. This differing tendency is attributed to the drawback of the AMIP experimental design. Over the summer monsoon oceans, SST anomalies are, to a large degree, induced by the atmospheric anomalies, thus enhanced rainfall tends to lower SST (negative correlation). In AMIP runs, however, the atmosphere is rendered only to passively respond to SST anomalies, thus positive SST anomalies enhance local rainfall (positive correlation). Uncovering the nature of the monsoon-ocean interaction may hold a key to advancing our understanding of the monsoon variability. It is also found that the intensity scores of simulating A-AM rainfall improve for those models that simulate better strength of rainfall anomalies in the El Nino region.