The work aims to ensure that fresh agricultural products can be distributed from the distribution center to the customer site in a short time, improve the distribution efficiency and reduce the distribution cost. With comprehensively consideration to the influence of temperature change on the fresh decay rate during unloading, the carbon emission cost during transportation and refrigeration, and the cost of violating the time window, a minimum distribution cost model was constructed. The carbon tax mechanism was introduced to analyze the carbon emission cost quantitatively. The optimization genetic algorithm was used to solve the problem to analyze the sensitivity of carbon emission price. The results showed that the optimized distribution cost was reduced by 3.8%. Compared with that before optimization, the carbon emission cost was reduced by 27.8%. The delivery time was 3.3% shorter than that before optimization. The optimized genetic algorithm is effective in reducing distribution cost and carbon emission cost. Distribution cost and carbon emission cost can be reduced through reasonable control of carbon emission price and other means.