A model is represented for deriving analytically the optimal lot size when a percentage of the produced items is defective. It is shown that the total unit cost increases as the defective percentage and/or the defective Denaltv cost increase. The effect of learning is also studied and it is shown that the total unit cost decreases as the learning level increases due to the negative effect of learning on both the actual defective percentage and the unit production time. It is also shown that anticipated total unit cost increases, due to increasing theoretical defective percentages, can be almost eliminated when high levels of learning are applicable. Finally the model is extended to the multiple item case with limited resources.