A spccific deficit in Theory of Mind (ToM) abilities has been proposed to explain the distinct pattern of sociallcognitivc, dcricits in individuals with autism (Frith, 1989). Leslie and Frith (1988), Pcmcr, Frith, Leslie and Lcckharn (1989) and Baron-Colien and Goodhart (1994) have also indicated that children with autism have a specific difficulty in understanding the principle that "seeing leads to knowing". However, other studies designed to measure ToM abilities have reported weak mcntalising skills in individuals with developmental language disorders (Shields, Varicy, Broks and Simpson, 1996), dcaf people (Peterson and Siegal, 1995) and individuals with intellectual difficultics (Shulman and Pilowsky, 1996). This study was designed to examine the developmental and cognitive correlates of one aspect of ToM. A method derived from ONeill and Gopnik (199 1) %vas adapted and children with autism (n = 17), children with intellectual difficulties of undifferentiated actiology (n = 19) and normally developing children (n = 21) were compared on their ability to understand the principle that "SEEING, FEELING and TELLING" lead to knowing. As predicted, individuals with autism (88%) and children with intellectual difficulties (63%) had difficulty identifying the role of informational access in knowledge, whereas the normally developing group did not (14%). Logistic regression analyses revealed that group membership and verbal mental age, when measured by the Test for Reception of Grammar (TROG; Bishop, 1983), were the best predictors of task performance. Discussion focuses on the implications of the findings and emphasises the need to carefully consider what tasks actually measure as well as the specific actiology of comparison groups when studying abilities and impain-nents of individuals with autism and intellectual difficulties