In this paper, multiagent genetic algorithm (MAGA) is firstly applied to tackle the synthesis of conformal sparse array, a constrained multi-objective optimization problem. Moreover, a model considered low peak sidelobe level (PSLL) is given for conformal sparse array synthesis. For the antenna array deployed on a quadric surface, the PSLL can be reduced by obtaining the optimal antenna element arrangement. An example of 256-element array synthesis with a 56% sparse rate proves MAGA as an effective optimization tool for conformal sparse arrays in low computational cost.