A linear Berea sand pack, initially containing three phases -- oil, water and air - was used to simulate experimentally the processes in the zones from the water bank through the evaporation and visbreaking region. Simulations using such process parameters as oil composition, system pressure, parameters as oil composition, system pressure, and gas injection rate were conducted. Every simulation followed programmed ascending-temperature steps with each step being approximately isothermal. The relative importance of each mechanism causing nonisothermal liquid movement was quantitatively determined and liquid saturation distributions were obtained. Results of the experiments showed that gas stripping was one of the main mechanisms responsible for oil flow in the hot water bank and that steam distillation was the chief mechanism for the oil movement in the steam plateau. Both mechanisms were primarily controlled by oil composition, system pressure and gas injection rate. The oil- and water-saturation distributions were reconstructed from the experimental results. A first-order equation approximated the oil saturation in the regions simulated and a second-order equation approximated the water-saturation distribution. Introduction Laboratory experiments, mathematical models, and field tests have been widely used to investigate the in-situ combustion process. Invariably, the results of these investigations have confirmed that the complex in-situ process possesses a number of distinct transient regions of varying physical and chemical importance. physical and chemical importance. Mechanisms associated with each of these regions have been qualitatively described. In general, these descriptions are based on gross experimental observations such as production history, temperature profile, produced fluid properties, etc. These qualitative descriptions of properties, etc. These qualitative descriptions of in-situ mechanisms are principally sound, but the scarcity of quantitative demonstration of nonisothermal liquid behavior, liquid-saturation distribution and process parameter effects on the mechanisms demands further investigation. This paper attempts to provide some quantitative information on several important mechanisms associated with in-situ combustion. A simple linear laboratory model is used to simulate the processes in the regions, starting from the water bank and ending at the evaporation and visbreaking region, which immediately precede the cracking region and combustion front. The experimental results give oil- and water-saturation distributions which should be useful in mathematical simulation and oil recovery estimation. REGIONS OF IN-SITU COMBUSTION Kuhn and Koch observed at least four coexistent transient regions in their experimental work using linear combustion tubes. Tadema performed in-situ combustion in a glass tube and described the mechanisms that would take place in each of these transient regions. Wu estimated the relative length of these regions in a combustion tube process under various experimental conditions. process under various experimental conditions. A schematic diagram of these regions is shown in Fig. 1. These regions were discretely reconstructed based on the description of Tadema and liquid production and temperature histories at the outlet end of a typical combustion tube experiment (performed by the Gulf Research and Development Co.). Starting from the injection end, these regions may be designated as the burned zone, combustion zone, cracking region, evaporation and visbreaking region, steam plateau, water bank, oil bank and initial zone. Each region has definite temperature and fluid-saturation characteristics. SPEJ P. 38