A suite of petroleum reservoir extracts from the Liaohe basin, NE China, was analyzed to investigate the effects andcontrols of biodegradation on petroleum composition.Systematic changes, seen as marked gradients in petroleum bulkcomposition, component concentrations and molecular indicators, have been observed in two biodegraded oil columns.In the Es3 column (1700-1850 m), biodegradation has removed substantial amounts of normal and isoprenoid alkanes.More severe biodegradation occurs in the shallower Es1 column (1500-1650m), with both steranes and hopanes beingpartially removed.Temperature has an overall control on biodegradation in the two columns, a lower reservoir temperatureleading to a higher degree of biodegradation.However, observed variation in the degree of biodegradation inthe Es3 column at different sites is controlled by water leg size, with a higher level of degradation being associated witha thicker water leg.The supply of nutrients from the water leg is thought to have a significant impact upon the degreeof biodegradation.In addition to water leg size/nutrient supply, the compositional gradients also are controlled by therelative rate of mixing of fresh oil charge.Gradients can only be conserved in biodegraded petroleum columns ifthe rate of diffusive mixing is similar to or lower than the rate at which the hydrocarbons are removed.A biodegradationmodel is proposed that couples geochemical and geological factors to provide a coherent approach forassessing the impact of degradation on petroleum.Our geochemical study indicates that biodegradation occurs mainlywithin a narrow region at or near the base of the oil column.The stable compositional gradients are produced after aninitial induction period involving diffusive transport of alkanes to the oil-water contact (OWC) and diffusion of biodegradationproducts such as 25-norhopanes and other metabolites away from the reaction site.The mixing of continuously-charged oil with 'residential' biodegraded oil by diffusion may be considered as one of the most importantfactors controlling the biodegradation process.The conceptual model illustrates all of these processes and opens thepossibility of model-driven prediction of oil properties and 'sweetspot' location in reservoirs.