The time-cost trade-off problem (TCTP) is an important branch in the project scheduling problem. However, the duration and cost of each activity could change stochastically as a result of uncertain factors. To meet the needs of real projects, an improved approach based on trapezoid fuzzy numbers is applied to estimate the uncertainty of time and cost. And then α-cut method is applied to decide the risk level. Furthermore, improved crossover and mutation methods for multi-objective genetic algorithm (MOGA) are used to make a large-scale computation possible. The efficiency of the proposed approach is verified by comparison with previous researches. In addition, economic analysis skill of finance cost is integrated into the new model to provide greater flexibility to managers when making decisions. Finally, time-cost tables under different risk levels for case examples are given and the advantages are investigated based on computation results.