Abstract： In order to explore the effects of machine failures on JIT delivery in pull environment, a combination model was proposed to jointly determine PM planning and production scheduling of a single machine. A new earliness/tardiness penalty membership function was proposed, and two kinds of PM—simple repair and replacement were considered. The earliness/tardiness penalty cost with due window and maintenance cost were included in the objective function. A new algorithm named GALSA based on GA was proposed to optimize the model. The proprosed model was compared with the enumeration algorithm and genetic algorithm to show its effectiveness. Computational experiments show the performance of this integrated model is satisfactory, compared with the conventional independent decision model and integrted model with single PM mode.