It is shown that the conventional Smith dead-time compensator can be modified in a fashion which leads to significant improvements in its regulatory capabilities for measurable disturbances or inputs. The proposed modification utilizes effectively the existing prediction capabilities of the conventional dead-time compensator and eliminates the need for a separate feedforward controller in many circumstances under which it is normally employed. The modified dead-time compensator possesses simultaneous feedforward and feedback actions. An additional tuning parameter which controls the degree of dead-time cancellation allows the designer to shape the response according to requirements. Illustrative examples show the advantages of the modified scheme.