In an effort to account quantitatively for performance differences between study‐test and anticipation methods, a rudimentary theory, the retention interval model, is proposed. This model postulates three different theoretical functions for the three operationally distinguishable and qualitatively different events in the retention interval: intercycle rest, intervening‐study‐events and intervening‐test‐events intervals. The retention interval model satisfactorily accounted for 144 data points based on 35, 136 observations (164 subjects) by using only 30 estimated parameters in three experiments (48 points via 10 parameters per experiment), with massed as well as spaced practice, under both anticipation and study‐test (RT) methods. In spite of a restrictive parameter estimation mode, close agreements were achieved between the model and data. The retention interval appears to generate powerful effects in determining the absolute performance levels for both anticipation and study‐test learning procedures in the various intercy