Problem-behavior resurgence can occur during lapses in the fidelity of interventions like functional communication training (FCT). Basic and translational research suggests that 1 potential strategy for mitigating relapse resulting from fidelity errors is to serially train multiple alternative responses. Two studies that evaluated serial training for this purpose (Lambert, Bloom, Samaha, & Dayton, 2017;Lambert, Bloom, Samaha, Dayton, & Rodewald, 2015) found discrepant results, with the desired recency effect (i.e., a greater magnitude of resurgence of recently trained alternative behavior, relative to resurgence of target behavior) only observed in the original study (Lambert et al., 2015). One hypothesis for discrepant results was that duration of exposure to baseline conditions differed considerably across studies (i.e., the former shaped new and arbitrary responses for analysis, and the latter addressed challenging behavior already at strength at study onset). Targeting arbitrary responses, we exposed 3 adult humans with developmental disabilities to 2 identically implemented analogues to serial FCT in a 2-component multiple schedule. Long-history baselines consisted of hundreds of sessions across months of implementation. Short-history baselines consisted of just a few sessions. During resurgence tests, we observed recency for 2 participants and primacy for 1. Thus, duration of exposure to baseline did not impact relative magnitude effects. Implications and future directions are discussed.