To address the stochastic discrete time-cost trade-off problems in construction projects， a double loop optimization procedure， which utilizes Genetic algorithm and monte Carlo simulation respectively in the outer and inner loops， is usually implemented. To reduce computational resources in the inner loop， an effective and dynamic strategy for allocating computational resources is proposed based on the statistical properties of the Monte Carlo simulation estimator. As shown by an illustrative example， the improved optimization algorithm can efficiently and stably solve the stochastic discrete time-cost trade-off problems in construction projects.