The response of the Los Alamos atmospheric general circulation model to the smoke from the Kuwaiti oil fires set in 1991 is examined. The model has an interactive soot transport module that uses a Lagrangian tracer particle scheme. The statistical significance of the results is evaluated using a methodology based on the classic Student's t test. Among various estimated smoke emission rates and associated visible absorption coefficients, the worst- and best-case scenarios are selected. In each of the scenarios, an ensemble of 10 30-day June simulations are conducted with the smoke and are compared to the same 10 June simulations without the smoke. The results of the worst-case scenario show that a statistically significant wave train pattern propagates estward-poleward downstream from the source. The signals favorably compare with the observed climate anomalies in summer 1991, albeit some possible El Ni¿o-Southern Oscillation effects were involved in the actual climate. The results of the best-case (i.e., least-impact) scenario show that the significance is rather small but that its general pattern is quite similar to that in the worst-case scenario. ¿ American Geophysical Union 1994