Achievement of specified performance tolerances despite plant parameter uncertainty is one of the main reasons for using feedback in control. The effect of sensor noise at the plant input is used as a measure of the resulting ‘cost of feedback’. This cost can be too high if only the plant output is available. Previous studies have shown how the cost can be greatly reduced if internal variables can also be used, even if all feedback loops must be returned only to the plant input. In plant-modification (PM) feedback, such loops are allowed to be returned also to internal plant variables, causing increase in plant signal levels needed to produce the specified system output. It is then necessary to specify the maximum increases allowed. In this paper, a complex five-loop PM system is illustrated to give more insight into PM system design. It is shown that the cost is then reduced significantly more than in the non-PM designs.