In the challenging context of heavy to extra heavy oil production, polymer flood technology appears to be a promising solution to enhance ultimate recovery of reservoirs. Several field applications have already shown the efficiency of such technologies, although the final incremental recovery and mechanisms involved are still poorly understood. Indeed, the characteristics of the viscous fingering effects that certainly play a role are rarely captured at the field scale or at the core scale. This work aims at comparing the results of two core experiments with polymer flood in secondary and tertiary mode, in reservoir conditions, in term of recovery as well as in terms of relative permeabilities. In both cases, experiments were carried out on reconstituted reservoir cores, with restored wettability, initially saturated with live oil partially degassed in a PVT cell to the expected pressure and viscosity at the start of the field test. Saturation profiles were measured with X-Ray scans; effluents were collected in test-tubes and analyzed by UV measurements. Additional follow-up with tracers was tested in order to better assess the breakthrough of different fluids as well as the polymer adsorption during the experiment. Although the viscosity ratio was still highly unfavorable, with a polymer bulk viscosity around 70 cP at 10s-1 and an oil viscosity estimated at 5500 cP, polymer floods exhibit an excellent recovery factor.