We examined the effects of baseline training sequence on the emergence of conditional discriminations in an intraverbal naming task. Thirty-two college students were randomly assigned to two groups. The tact-intraverbal (TI) group first learned to vocally tact eight visual stimuli using a unique verbal label for each stimulus, and then to intraverbally relate four pairs of verbal labels. The intraverbal-tact (IT) group received the same training but in the opposite sequence. Both groups then received a match-to-sample test involving the visual stimuli alone. On average, the TI group had significantly shorter reaction times than the IT group throughout all four test blocks, even when controlling for intraverbal retention, which was lower in the IT group. Accuracy on the MTS test did not differ significantly between groups when controlling for intraverbal retention. However, MTS accuracy and intraverbal retention were strongly correlated in the IT group but uncorrelated in the TI group. We suggest the effect of training sequence reflects different sources of stimulus control available to subjects in different groups when confronted with the novel MTS trials.