Abstract: Geological risk evaluation reveals the uncertainty of converting an anticipated resource potential in an identifi ed exploration target into an economic petroleum accumulation prior to drilling. Providing consistent and unbiased estimates of geological risk for all exploration targets in an area of interest is essential in risk evaluation. This paper discusses the potential use of an object-based stochastic procedure for geological risk evaluation at the play level. The object-based model of SuperSD, as a data integration tool, can integrate geological information of petroleum system elements that control occurrences of petroleum pools with the spatial correlation characteristics of the discoveries in a play. Other information, such as the estimated total play potential and size characteristics, from conventional resource assessments can be also incorporated into the model as geological constraints. The uncertainty associated with the data and domain knowledge in the predicted pool locations is expressed as a probability map, representing the geological exploration risk of the play. By incorporating all available information and checking the consistency of models with geological constraints, the object-based model improves predictions of petroleum occurrences, thus enhancing the associated risk evaluation. The proposed method was applied to the fractured petroleum play of the western Qaidam Basin in northwestern China in order to illustrate the use of the object-based method for risk evaluation.