In an experiment on 6 young adults advance (precued) information of the correct choice response was utilized completely: (a) For precue-to-stimulus intervals (PSIs) clearly shorter than the difference between mean choice and simple reaction time (RT), median response latency (L) measured from precue onset was invariant; (b) For clearly longer PSIs, median RT was very near the value for simple RT. This precue-utilization effect would be expected if response actualization had been delayed until the response had been selected and if the requirement for discrimination and selection had no adverse effect on readiness to respond. Donders' (1868/1969) hypothesis in his initial application of the subtraction method, that choice and simple reactions are identical except for the serial insertion of discrimination and selection operations in the former, is thereby strongly supported. If this formulation is accepted, models that hold that response processing can overlap other processing stages may be considered valid only for response selection, not response actualization.