Summary: We deal with the heuristic solution of the classical job shop problem. Both the constructive and the iterative phase of our algorithm apply insertion techniques combined with beam search. In the first phase we successively insert the operations into feasible partial schedules. In the iterative phase we generate paths in a particular neighbourhood graph instead of investigating the neighbourhood completely. To select “interesting” neighbours, we use the combinatorial path structure of feasible solutions of the job shop problem. The results of our algorithm are compared with those from other well-known methods on benchmark problems.