We have simulated the CO2 seasonal cycle of the Martian atmosphere and surface with a hybrid energy balance model that incorporates dynamical and radiation information from a large number of general circulation model (GCM) runs. This information includes heating due to atmospheric heat advection, the seasonally varying ratio of the surface pressure at the two Viking landing sites to the globally averaged pressure (rk), the rate of CO2 condensation in the atmosphere, and solar heating of the atmosphere and surface. The GCM runs collectively covered a full set of seasonal dates and a large range of dust optical depths. We have compared the predictions of the energy balance model with the seasonal pressure variations measured at the two Viking landing (VL)sites and the springtime retreat of the seasonal polar cap boundaries. Numerical experiments with the energy balance model indicate that the following quantities have a strong influence on the VL seasonal pressures: albedo Ais of the seasonal CO2 ice deposits, emissivity eis of this deposit, atmospheric heat advection, and the pressure ratio rk. This last factor does not enter into the seasonal CO2 condensation/sublimation cycle in a significant way. The numerical experiments also indicate that the following factors have only a minor effect on the VL pressures: (1) the net radiative effects (solar plus thermal) of atmospheric dust at the latitudes of the polar caps, and (2) the subsurface heat conduction. The significant influence of the pressure ratio rk on the VL seasonal pressures is due to large seasonal variations in the global distribution of surface pressure. At low and mid-latitudes, these “weather” variations are engendered by seasonal changes in the Hadley circulation and by seasonal changes in the atmospheric scale height close to the surface. Comparison of the VL1 and VL2 pressures with one another provide direct evidence for the presence of such a “weather component” in the measured pressures. The differential weather component (VL2-VL1) derived from the data is reproduced approximately by the energy balance model. We find that the seasonal weather variations account for about 20% and 30% of the seasonal pressure variations measured at VL1 and VL2, respectively, that dynamical and scale height variations make comparable contributions to the weather component during years without global dust storms, and that the dynamical contribution is the larger one during years with global dust storms. Interannual variations in the weather component, rather than variations in CO2 condensation rates, are the dominant sources of the observed interannual variations of pressure during the season of global dust storms. Optimum fits to the Viking pressure measurements and the data on the polar cap boundaries are achieved with values of about 0.45 and 0.75 for Ais and eis, respectively. The former value is consistent with available photometric determinations of the albedo of the seasonal caps, while the latter value, especially in light of infrared thermal mapper brightness temperatures at high latitudes, may reflect, in part, the influence of the polar hood on the radiation balance of the winter polar regions.