Alkaline-cosolvent-polymer (ACP) flooding of a heavy acidic oil (320 cp) was studied in a two-dimensional (2D) sand pack. An ACP formulation was developed by studying the phase behavior of the oil with several alkaline-cosolvent formulations. The effectiveness of the ACP formulation was validated in a 1D sand pack by conducting a waterflood followed by a stable ACP flood. Sand was then packed into a 2D square steel cell equivalent to a quarter five-spot pattern. Several ACP floods were then conducted in this 2D cell to study both the displacement and sweep efficiency of ACP floods. First, the polymer concentration in the ACP formulation was varied to find an optimum ACP slug viscosity where the oil recovery and pressure gradients are acceptable. When the ACP slug viscosity was about 1/4 of the oil viscosity, oil recovery was 76% OOIP; we adopted this viscosity as the optimum viscosity. Then, using this ACP slug viscosity, the waterflood extent, after which the ACP flood was initiated, was varied (0-1 pore volume). The oil recovery, oil cut, effluent concentration and pressure drop were monitored during the floods. The timing of the tertiary flood did not change the tertiary oil recovery very much. However, shorter the waterflood, faster was the oil recovery under the constant injection rate constraint. Sweep efficiency during secondary ACP flood was lower than that in the tertiary ACP injection. Gravity effect on sweep was significant under ultralow IFT.