Directional response biasing as a function of the recency of an interpolated act was examined on a lever device in a within-S design. The temporal occurrence of an interpolated act was systematically varied (5, 30, or 55 sec. after the criterion act) for three retention intervals (15, 40, and 65 sec.), and the location of the interpolated movement was held constant ± 40 deg. from the criterion position. Analysis of constant errors showed significant positive directional biasing for interpolated acts greater than criterion positions. Further analyses at the 40-and 65-sec. retention intervals supported a recency interpretation. As the retention interval increased, the positive constant error was maintained for interpolated acts with the same post-biasing interval, whereas constant error became increasingly negative as the post-biasing interval increased. Examination of variable error revealed a slight increase over time and no significant differences between directional conditions. The recency findings were discussed in terms of the view that reproduction of a criterion act is dependent on the relative decay states between the interpolated and criterion traces.