This thesis presents a heuristic algorithm for multi-item multi-level dynamic lot sizing problem under multiple bottleneck facilities. The problem is formulated as a mixed integer programming model. Subproblems for each item is obtained by taking Lagrangean Relaxation at each stage through bill of material sequence. Each subproblem considers the rest of the system like a master problem. And marginal analysis is used for updating multipliers. And solution procedure is presented. Finally, computational results and suggestions for further study are presented.