A novel combination of genetic algorithms and constraint satisfaction modelling for the solution of two and multi-layer over-thecell channel routing problems is presented. The two major objectives of the optimization task are to find an optimal assignment of nets to over-the-cell and within the channel tracks, and to minimize the channel widths through a simple but effective iterative routing methodology. Two genetic algorithms cooperate in a nested manner to perform the optimization task. The results obtained using the benchmark problems published in literature indicate that, without any predefined fixed upper/lower channel widths, the implemented algorithm outperforms wellknown channel routers.