Summary Oil and water production data are regularly measured in oilfield operations and vary from well to well and change with time. Theoretical models are often used to establish the production expectation for different recovery processes. A performance surveillance understanding can be developed by comparing the field production data with the production expectation. This comparison generates quantitative or qualitative signals to determine whether the producer meets production expectations or the producer is underperforming and appropriate operational action is required to address the underperformance. The case study is for the South Belridge diatomite in California. This hydraulically fractured diatomite reservoir is currently under waterflood and steamflood. A methodology is proposed to establish the production expectation from historical production data. For primary depletion, the formation linear and bilinear flow models are applied to producers with vertical hydraulic fractures. For waterflood, an analytical method derived from the Buckley-Leverett displacement theory is used. Those analytical methods can predict production and provide surveillance signals for producers in the primary and waterflood recovery stages. For steamflood, a semiquantitative performance/surveillance criterion is proposed on the basis of understanding the mechanistic oil banking concept and reservoir simulation results for steamflood and waterflood. With those models representing expected production performance, an integrated flow regime diagram is proposed for production surveillance. A performance expectation can be developed for an individual producer. A significant overperformance relative to the expectation normally indicates changes in the recovery mechanism or improvement in sweep efficiency. A significant underperformance usually signifies an operational issue that requires correction to optimize the production performance. In the case study, the surveillance methodology for producers under primary depletion, waterflood, or steamflood is demonstrated by use of historical production data. In addition, water channeling between injectors and producers and its impact on production performance are discussed. On the basis of this surveillance methodology, some operational actions were proposed, and successful results are demonstrated. Examples of forecast for an individual producer in the primary depletion stage and field scale prediction in the waterflood stage are provided. Application indicates that the proposed methodology can serve as a convenient and practical tool for reservoir surveillance and operational optimization.