Foam-like dispersions of CO2 into brines can reduce the mobility of drive fluids in CO2 floods. To evaluate the effectiveness of such foam-like dispersions, time-consuming laboratory coreflood tests are routinely used. Because of the costliness of such coreflood tests, simple qualitative tests have long been employed to screen potential surfactants. Then only a few of the better candidates are subsequently evaluated in coreflood tests. There are a number of disadvantages of such qualitative tests; therefore we developed, instead, a quantitative screening process. Our quantitative process is based on two simple, quick laboratory tests and a neural network interpretation of the test data. The neural network predicted CO2 mobility reduction values which correlated well with the mobility reductions seen in coreflood tests.