The mixed refrigerant (MR) process is widely used for the liquefaction of natural gas (NG). The determination of MR composition with respect to variations in ambient conditions is crucial but difficult. PRICO is one of the simplest MR processes, and it is frequently used in small and mid-scale natural gas liquefaction plants. However, the determination of MR composition in the PRICO process is not easy. In this work, we presented a potential solution to this problem. First, operating pressures in the PRICO process were assumed to be fixed. We then programmed a genetic algorithm (GA) method coupling the process simulation software Aspen Plus and optimised the MR composition solution under different cold box inlet temperatures. Linear regression was performed on the MR composition to derive a set of functions, which were then validated for feasibility and energy efficiency. The results show that when the ambient temperature increases, the concentrations of methane, ethylene and propane should decrease, and isopentane should increase. This work may be helpful for designers and operators working with the PRICO process. (c) 2012 Elsevier Ltd. All rights reserved.