This paper investigates the existence and position of virtual targets during the production of stop consonants. Using the equations from general Tau theory to model the time-course of tongue constriction formation movements, targets were estimated by fitting these equations on observed tongue constriction variables extracted from real EMA data from 2 native speakers of English. Results suggest that targets are virtual for 50 to 60% of movements. For these movements, virtual targets of the tongue tip constriction are predicted to occur around 0.1 cm beyond the palate, and virtual targets for the tongue dorsum constriction are predicted to occur between 0.05 and 0.2 cm beyond the palate. Our results suggest that the time-course of movement is planned so that the onset of closure occurs with relatively high velocity: closure onset is generally located very close in time to the time of peak velocity.