Abstract: We construct an optimization model with two-tier structure to quantify the schedule risk control of IT outsourcing projects based on the principal-agent model. We design an adaptive tabu search algorithm to solve the model. The improved algorithm combines diversified search mechanism with loop execution of tabu search. The intensity of tabu search and the strength of diversified search are automatically adjusted according to the feedback of fitness value. While, the improved algorithm applies the greedy strategy to construct the initial solution. Two neighborhood structures are used crosswise to improve algorithm optimization ability. The experimental results show that schedule risk control significantly reduces the risk of delays in IT outsourcing projects, and maximizesthe benefits of both the principal and the agent. Compared with the genetic algorithm, simulated annealing algorithm, basic tabu search algorithm, adaptivegenetic algorithm and adaptive simulated annealing, the improved algorithm is obviously superior to other algorithms in terms of convergence degree and stability. As the scale of the problem increases, the superiority of algorithm is more obvious.