The present series of 4 experiments investigated the probability of responding in accord with equivalence in adult human participants as a function of increasing or decreasing delays in a many-to-one (MTO) or comparison-as-node and one-to-many (OTM) or sample-as-node conditional discrimination procedure. In Experiment 1, 12 participants started with simultaneous matching, followed by 0-s, 2-s, and 4-s delayed matching, while the other 12 participants started with 4-s delayed matching, followed by the decreasing delays in a MTO training structure. Results suggested that the likelihood of responding in accord with equivalence is enhanced as a function of increasing delays during training, and also that starting with longer delays increases the probability of responding in accord with equivalence in simultaneous matching. In Experiment 2, 6 participants started with simultaneous matching, followed by 0-s, 2-s, and 4-s delayed matching in a OTM training structure. Results showed that 6 of 6 participants responded in accord with equivalence. In Experiment 3, 6 participants started with simultaneous matching, followed by 0-s, 3-s, and 9-s delayed matching in a OTM training structure. Results showed that 6 of 6 participants responded in accord with equivalence. In Experiment 4, 6 participants started with simultaneous matching, followed by 0-s and 3-s delayed matching with a restriction to engage in mediating behavior during testing. Results showed that none of the 6 participants responded in accord with equivalence during 3-s delay. It is argued that if delayed matching requires (or evokes) precurrent behavior (responses that increase the effectiveness of a subsequent current behavior in obtaining a reinforcer), precurrent behavior may in turn facilitate responding in accord with equivalence. Higher reaction times during initial testing may indicate precurrent problem-solving behavior prior to the selection of a comparison stimulus. (Contains 7 tables and 10 figures.)