The paid use and trading of emission permits system are a major mechanistic innovation in the field of environmental resources management in China. In this system, manufacturers must purchase emission permits on the primary market under the precondition that the purchase ceiling and price will be determined by the government; subsequently, they may trade emission permits with other manufacturers on the secondary market subject to established rules. This inevitably leads to uncertainty regarding the emissions trading price on the secondary market, while the manufacturer is making the initial purchase decision; it highlights the need to investigate optimal purchase planning solutions under conditions of uncertainty from the perspective of production optimisation. This paper examines the manufacturer's random profit function, as well as its mean and variance, under uncertain emission trading prices. Using the mean–variance framework, an optimisation model with differentiated decision objectives is established. Accordingly, the manufacturer's optimal initial purchase planning approach is demonstrated with respect to each decision objective. The conclusions demonstrate that when the manufacturer exhibits risk preference, the optimal purchase planning process may systematically deviate from the risk‐neutral approach.