Cyclic solvent injection (CSI) process has been suggested as the most effective and promising solvent-based method in some heavy oil reservoirs in which thermal-based methods are not suitable due to economic constraints and environmental concerns. In solvent-based methods, many studies indicated that pure propane can more significantly reduce the viscosity of crude heavy oil than methane or carbon-dioxide, especially when the propane is injected with gaseous phase under the pressure slightly lower than its dew point. Actually in field application, to achieve a great effect of viscosity reduction with gaseous solvent, the mixture solvent rather than pure propane is injected to guarantee the saturation pressure of injected solvent higher than the reservoir pressure. However, most current studies have been done only by the pure solvent for investigating mechanisms of the CSI process. Therefore, it's very necessary to conduct further investigation on CSI process by the mixture solvent. In this study, three CSI tests by the mixture solvent (60% mol. CO2 and 40% mol. C3H8) with the decline rate of 12.5 kPa/min, 5.0 kPa/min and 1.0 kPa/min respectively were conducted in a sand-pack model with the diameter of 15.24 cm and the length of 30.48 cm. Three pure propane CSI tests from previous study, also with the decline rate of 12.5 kPa/min, 5.0 kPa/min and 1.0 kPa/min respectively, were used to perform comparison analyses with three mixture solvent CSI tests for better understanding driving mechanisms of the CSI process. Comparison results show that pure propane tests had larger recovery factor and average oil production per single cycle than mixture solvent tests. The amount of the propane diffusing into the heavy oil in mixture solvent tests was lower than that in pure propane tests because of the negative effects of the carbon-dioxide. The asphaltene precipitation and production time significantly impacted the recovery factor of diluted oil in mixture solvent tests. It's indicated that the effects of the decline rate on the asphaltene precipitation were more sensitive in mixture solvent tests. As well as in pure propane tests, the decline rate can also be optimized in mixture solvent tests. It's suggested that 5.0 kPa/min is the optimal decline rate with the largest recovery factor and average production per single cycle in mixture solvent tests. In addition, hysteresis of the declined pressure of the sand pack model led to a high pressure difference and intensive oil production between 300 kPa and 200 kPa in the 1.0 kPa/min mixture solvent test