From 1986 to 1996, oil recovery in the US by gas injection increased almost threefold, to 300,000 bbl/day. Carbon dioxide (C02) injection projects make up three-quarters of the 191,139 bbl/day production increase. This document reports experimental and modeling research in three areas that is increasing the number of reservoirs in which CO2 can profitably enhance oil recovery: 1) foams for selective mobility reduction (SMR) in heterogeneous reservoirs, 2) reduction of the amount of CO2 required in CO2 floods, and 3) low interfacial tension (1FT) processes and the possibility of CO2 flooding in naturally fractured reservoirs. CO2 injection under miscible conditions can effectively displace oil, but due to differences in density and viscosity the mobility of CO2 is higher than either oil or water. High CO2 mobility causes injection gas to finger through a reservoir, causing such problems as early gas breakthrough, high gas production rates, excessive injection gas recycling, and bypassing of much of the reservoir oil. These adverse effects are exacerbated by increased reservoir heterogeneity, reaching an extreme in naturally fractured reservoirs. Thus, many highly heterogeneous reservoirs have not been considered for CO2 injection or have had disappointing recoveries. One example is the heterogeneous Spraberry trend in west Texas, where only 10% of its ten billion barrels of original oil in place (OOIP) are recoverable by conventional methods. CO2 mobility can be reduced by injecting water (brine) alternated with CO2 (WAG) and then further reduced by adding foaming agents-surfactants. In Task 1, we studied a unique foam property, selective mobility reduction (SMR) , that effectively reduces the effects of reservoir heterogeneity. Selective mobility reduction creates a more uniform displacement by decreasing CO2 mobility in higher permeability zones more than in lower permeability zones. We show that SMR varies with the type of surfactant, the concentration of surfactant, and injection rate, and that SMR has been demonstrated in multiple permeability zones, with and without capillary contact. The most effective surfactant concentrations for foaming are fairly low, below the critical micelle concentration. To rapidly identify and develop less expensive foaming agents, a screening method to identify surfactants with good SMR potential has been developed. In Task 2, a model required for reservoir performance predictions was developed for foam mechanisms and properties, and incorporated into a fully compositional simulator and a pseudomiscible reservoir simulator. Reservoir core studies were performed to aid in the development of the foam model. The effects of flow rate, surfactant concentration, foam quality (flow fraction of CO2 versus liquid), and rock permeability and composition on foam have all been determined. The model has been tested using a scheme selected to mimic the CO2-foam pilot test performed in an earlier project, "Field Verification of CO2-Foam.''The results show a decrease in the gas production rate, an increase in oil production, and significant sweep improvement. They are consistent with pilot test results. Many CO2 injection projects maintain reservoir pressures well above the minimal miscibility pressure (MMP) and, as a result, require more purchased CO2 and additional compression cost. Thus, this project includes studies of the effects of pressure, temperature, and oil composition on fluid behavior and oil recovery. Results indicate that lowering the pressure will have minimal effects on recovery as long as miscibility is maintained. Related to the fluid study is the development of a phase behavior-preprocessor for reservoir simulations and other fluid property prediction methods. Naturally fractured reservoirs do not meet classic CO2 injection project screening criteria due to the expectation of excessive channeling of low viscosity CO2 through natural fractures. However, in Task 3, our laboratory results indicate that CO2 injection can produce significant oil production by gravity drainage in naturally fractured reservoirs if the fractures have sufficient vertical relief and significant ,density. An important accomplishment, therefore, was the development of a new mathematical model based on Darcy's law and film flow theory to describe free-fall gravity drainage with equilibrium fluids. This research is aiding in the development of CO2 flooding procedures for the Spraberry DOE/Industrial Class ill reservoir project. Multiphase flow behavior in fractured reservoirs was also investigated in Task 3. Understanding the relationship of fluid flow and reservoir heterogeneity in fractured reservoirs is a key factor in developing a strategy of improving oil recovery in these reservoirs. The ability to measure and predict 1FT under reservoir conditions and to describe gravity drainage are necessary developments toward the goal of improving oil recovery in fractured systems that previously have not been seriously considered for CO2 flooding. To this end, an apparatus for measuring 1FT at reservoir conditions was designed, constructed, and operated. The traditional method for determining 1FT using the pendant drop technique does not work well at low IFf. A new method for determination of IFf was developed and shown to be effective at low IFf. The fourth Task of this project was the transfer of information to the public domain. Besides the required quarterly, yearly, and final report, we have been aggressive in presenting and publishing. To date, 16 papers are in print and several have been accepted for upcoming events. We also organized two CO2 Oil Recovery Forums; the first, October 23-24, 1996, and the second, October 29-30, 1997. Both were well-attended.