School transport is essential to ensure access and permanence of students in public schools, especially in rural areas, where students are located in a large area with low density and roads are in precarious conditions. This work aims to apply the Iterated Local Search metaheuristic to route 13.664 rural students in the state of Rio de Janeiro. For this, we considered the School Bus Routing Problem (SBRP), with heterogeneous fleet and single school, in order to minimize the total cost considering the vehicles capacity constraints and maximum travel distance. To apply the method, the data provided by Secretaria de Estado de Educação do Rio de Janeiro (SEEDUCRJ) were considered. The results are presented in two scenarios, the first considers data from 79 routes used by SEEDUC-RJ to compare the results obtained with the ILS. The method showed a reduction of 40.5 percent in the average cost of routes and 46 percent in the average mileage per student. The second scenario considers the routing of all students, who were divided into 506 instances considering school and shift. The largest routed instance has 534 students. The results consolidated by municipality are presented and show the concentration of municipalities with the highest average cost per route in northwestern Rio de Janeiro. The implementation of the proposed routes can bring significant savings with expenses related to rural school transport, in addition to indicating an increase in the level of service for students, with a reduction in the average mileage per student.