The in situ forward combustion process in a Utah tar sand was evaluated at alternative operating conditions. Five bench-scale tube reactor experiments were conducted to study the effects of the injected air flux on the oil yield, product oil quality, and process efficiency. Air injection fluxes ranged from 16 to 111 scfh/ft/sup 2/ and one test was conducted using steam dilution of the injected air. Regardless of the injected air flux, the oxygen consumption approached 100% and the combustion process became self-sustaining at ignition temperatures around 650 to 750/sup 0/F (343 to 399/sup 0/C). After ignition, all of the oxygen reacted with residual coke and the combustion temperatures increased to 1500 to 1900/sup 0/F (815 to 1035/sup 0/C). Residual coke was the principal source of fuel for the combustion tests since the available coke, as determined by Fischer assay (36 to 39 wt.% initial bitumen), exceeded the fuel required for combustion. Other factors also indicated residual coke was the principal fuel: (1) oil production ceased before oxygen breakthrough, (2) unburned coke was measured on the lowermost spent sand, and (3) the calculated equivalent H/C ratio of the fuel burned was low. Although there was only a small increase in the weight yield of the product oils (38 to 43 wt.%), the volumetric oil yields (40 to 49 vol.%) increased with decreasing gas fluxes largely because of product oil upgrading. The product oil quality improved with decreasing gas flux probably because of the increased residence time for the pyrolysis of undistilled bitumen. At low gas fluxes, the bitumen distillation rate was limited by vapor-liquid equilibrium and more undistilled bitumen was pyrolyzed. Increased pyrolysis of this bitumen reduced the molecular weight of the product oil. 10 refs., 4 figs., 4 tabs.