Background: It is not known whether those with co-morbid fragile X syndrome (FXS) and autism represent a distinct subtype of FXS; whether the especially severe cognitive delays seen in studies of young children with co-morbid FXS and autism compared with those with only FXS continue into adolescence and young adulthood; and whether autism in those with FXS is "true autism", i.e. reflects the same underlying problems as idiopathic autism. Method: We compared the non-verbal IQ of adolescents and young adults with co-morbid FXS and autism ("n" = 10) with those with only FXS ("n" = 44). We then created a subsample of those with FXS only, matched on non-verbal IQ, mental age and gender ("n" = 21) to the subsample of those with co-morbid FXS and autism. We compared the two groups on measures of expressive language, receptive language (lexical, grammatical morphology and syntactic patterns), and a theory of mind task. Results: Those with co-morbid FXS and autism had lower non-verbal IQs than those with only FXS. The participants with co-morbid FXS and autism did not perform as well as the cognitive ability- and gender-matched participants with only FXS on the three measures of receptive language or the theory of mind task; there were no differences on the expressive language measure. Conclusions: Our findings support the notion that those with co-morbid FXS and autism represent a distinct subtype of FXS, with more impairment in receptive language and theory of mind even when controlling for their lower non-verbal IQ relative to those with only FXS. The greater cognitive impairments observed in those with co-morbid FXS and autism continues into adolescence and young adulthood; and the autism seen in those with FXS appears to be the same as idiopathic autism.