ABSTRACT A semi-empirical method is proposed to forecast waterflood performance using classical waterflood models. The approach adopts an average flood pattern for waterflood performance projection. The pattern waterflood performance was obtained by using a semi-empirical approach based on a comparison of the computed results from Buckley-Leverett-Welge, Dykstra-Parsons, and Craig-Geffen-Morse methods. The approach allows the engineers to adjust the effects of. dominant factors such as displacement mechanisms, vertical variation of horizontal permeabilities, mobility ratio, and areal sweep efficiency. The technique also allows for fillup of initial gas saturation which can be adjusted to match the response time. The field-wide waterflood performance is computed using the average pattern waterflood performance and a prescribed pattern development schedule. The technique can be used for initial waterflood project evaluation or for matching the performance of a matured waterflood. The approach is practical and helpful for reservoir engineers working on waterflood project evaluation and management. The approach has many inherent limitations. There are numerous assumptions made in the Buckley-Leverett-Welge, Dykstra-Parson, and Craig-Geffen-Morse methods such as homogeneous reservoir layers with constant porosity, permeability, and initial saturation. The proposed approach does not provide individual well performance and does not give a unique answer to the waterflood performance forecast problem. Due to the inherent limitations, the reservoir engineers should exercise good reservoir engineering judgements when using the proposed waterflood calculation technique. A field example is used to illustrate the steps for completing the waterflood performance projection. A package of personal computer programs could easily be developed to relieve reservoir engineers from repetitive and laborious calculations and aid in parametric sensitivity study.