This paper is concerned with memory functions in sequentially structured behavior. Twenty-five 4- and 5-year-old preschool children participated in a prediction experiment in which a stack of cards (each card alternately having a patch of red or green tape on it) was displayed to the child. The child was presented with a card and asked to predict the color on the next card. Two interval lengths, a long and a short, were used between presentation and prediction. The subject's performance, it was thought, was affected by (1) memory of each trial, (2) effects of the previous response, (3) lagging of attention, (4) guessing, and (5) the variation in interval length. The results from 100 trials indicated that the probability of an error, given a correct response on the previous trial, is greater following the long interval than following the short. It was also found that the probability of an error, given a correct response, is less than or equal to both the probability of a correct response, given an error, and the probability of an error, given an error. The theoretical basis of this task is being used to develop a recall task similar in form to the prediction task. (WD)