To examine contrasting predictions between fuzzy trace theory (FTT) and activation monitoring theory (AMT), the current study examined the proportion of DRM false memories produced by children vs adults. In contrast to previous studies, backward associative strength (BAS) was well controlled, and the method ensured that subjects' attention was focused on lexical/semantic information. Specifically, DRM lists for children were constructed from child norms, and DRM lists for adults were based on adult norms. In addition, to ensure that attention was focused on lexical/semantic information, lists were read aloud to the subjects, and responses were made verbally. Children produced more false memories than adults, supporting AMT. Response modality was varied in a second Experiment. Children recalled more items when responding orally than when writing responses. Adults showed no effect of response modality. These results indicate that when BAS is controlled, and both children's and adults' attention is focused on semantic information, false memory occurs more frequently in children than adults. We propose that this pattern is indicative of a less developed attentional system in children, is consistent with activation monitoring theory, and suggests important boundary conditions for the occurrence of developmental reversals.