One-dimensional laboratory experiments were conducted to evaluate reverse combustion both as a recovery process and as a reservoir preheating mechanism before steam displacement. Pairs of similarly packed tubes were subjected to identical reverse combustion conditions, and one packed tube of each pair was subsequently steamflooded. Reverse combustion effectively preheated the tar sand without developing matrix plugging. Peak combustion temperatures increased with increased air flux and ranged from 600/sup 0/ to 1050/sup 0/F (316-566/sup 0/C). Oil yields averaged 33 wt. % of the initial bitumen and wre independent of injected air fluxes between 50 and 100 scfh/ft/sup 2/. The fuel consumption averaged 24 wt. % of the initial bitumen, and most of the remaining bitumen was converted into an unrecoverable coke (33 wt. % initial bitumen). Consequently, very little bitumen or oil (less than 10 wt. % initial bitumen) was available for recovery during the subsequent steamflood phase. Coking occurred at temperatures as low as 400-450/sup 0/F (200-230/sup 0/C), and tar sand ignition occurred at approximately 600/sup 0/F (315/sup 0/C). Low-temperature oxidation reactions were dominant prior to ignition and were characterized by low levels of carbon oxides in the product gas, increased apparent hydrogen/carbon ratio in the combustion products, increased water production, increased acidity in the product water, and increased oxygen in the product oil. The quality of the reverse combustion product oil was significantly better than the original bitumen. The API gravity increased from 10.6/sup 0/ (bitumen) to 21.2/sup 0/ API, and the viscosity at 140/sup 0/F (60/sup 0/C) decreased three orders of magnitude from 12,800 cp for the bitumen to 18 cp for the product oil. 12 refs., 5 figs., 6 tabs.