252 fifth-grade students, categorised into anxiety-ability levels, were exposed to one of three forms of pre-program experience in an attempt to manipulate achievement motivation: (a) a pre-test followed by a satisfactory report, (b) a pre-test followed by no report, and (c) a pre-test on material other than that covered by the program. After the pre-program experience students completed either a large or small step (SS) version of an English money program. Using four criteria (immediate post-test, delayed post-test, time, and program errors) ability was significant for all but time (p<0·01), anxiety was significant only for errors and only on the SS program (p<005), and the Treatment X Anxiety interaction was significant for the delayed post-test on the SS program (p<0·01). This interaction was such that high-anxious students performed better, without a pre-test while low-anxious students did better with a pre-test. Discussion centred on the explanation of this interaction and why the SS program produced more differentiation than the large step.