The scheduling problems have been analyzed considering that the processing time of operations is known and normally without maintenance activities and set up times between jobs. Minimizing makespan is the most studied criterion, which does not consider important aspects for measuring the customer service like the due date and the importance between customers. Besides, the few level of publications based on regular criteria have not considered the maintenance activities and sequence dependent set-up times, which affects the finalization of jobs. In this paper, we propose a hybrid approach for minimizing regular criteria in the Job-shop Scheduling problem with maintenance activities and sequence dependent set-up times. It is an ant colony Min-Max system, which is improved with a local search algorithm at increasing the neighborhood. Our approach makes use of the disjunctive graph model to represent schedules and support the search for an optimal at reversing a critical arc that affects the criterion during the improvement phase and pheromone is supplied to the arcs that solve the problem if the global optimal is gotten. In the diversification phase, a parallel search of k ants is executed considering the pheromone on the arcs to escape of a local optimal and the best ant returns to improvement step. The quality of our approach is illustrated on known instances at adding information. The superiority respect to local search process and a classic ACO is evaluated. Finally, a reference of results is proposed for various regular criteria.