This paper describes a delayed response experiment which tested children's memory for the location of a hidden object. Eight boys and eight girls at each of two age levels (25 and 37 months) were assigned to each of four experimental conditions and given eight trials in a 9-choice task. On each trial the child saw the object hidden in one of nine boxes, played with a toy for a 25-second delay period, then searched for the object. In one condition, only location cues were present and relevant for finding the object. In the other conditions, both location and picture cues were available. In these conditions, pictures were either: (1) redundant with location cues, so that both were relevant, (2) the solely relevant cues, or (3) irrelevant. Both age groups responded well on the basis of location cues alone. When picture cues were relevant and redundant with location, correct responses of 3-year-olds improved significantly, that of younger children only slightly. When picture cues alone were relevant, 3-year-olds performed well, but 2-year-olds did poorly. When picture cues were irrelevant and location relevant, the reverse occurred. These results indicate that there is a shift from reliance on location cues at age two to increased spontaneous utilization of pictorial cues by age three. (Author/BF)