In this paper, the Computational Fluid Dynamics (CFD) technique is used to investigate the influences of varying cooling loads on the ejector pressure recovery performance in an ejector-based multi-evaporator refrigeration system (EMERS) using R134a as the refrigerant. The performance of pressure recovery in the EMERS reflects the performance of the compression energy saving. The developed CFD model is first validated by actual experimental data from the EMERS. Turbulence model constants are carefully selected in order to minimize the model prediction error. Over 200 different cases are studied using the model to find the effects of varying cooling loads on pressure recovery ratio. The results indicate that pressure recovery ratio is very sensitive to the varying primary and secondary flow cooling loads. The maximum pressure recovery ratio can reach 60% as the cooling loads vary. It was found that in order to keep the system stable, the primary and secondary cooling loads should be maintained within ±5% and ±10%, respectively, in which case the pressure recovery ratio will have a maximum ratio of 32.8%.