It has been suggested that autism may arise as the result of exposure to high concentrations of prenatal testosterone. There is evidence that the ratio of the lengths of the 2nd and 4th digit (2D:4D) may be negatively correlated with prenatal testosterone. We measured 2D:4D in 95 families recruited via the National Autistic Society, UK. The sample comprised a total 72 children with autism (62 males, 10 females; age range 2 to 14 years), including 23 children (20 males, three females) with Asperger syndrome (AS), 34 siblings, 88 fathers, 88 mothers and sex- and age-matched control participants. We found that the 2D:4D ratios of children with autism, their siblings, fathers and mothers were lower than population normative values. Children with AS, who share the social and communicative symptoms of autism but have normal or even high IQ, had higher 2D:4D ratios than children with autism but lower ratios than population normative values. There were positive associations between 2D:4D ratios of children with autism and the ratios of their relatives. Children with autism had lower than expected 2D:4D ratios and children with AS higher ratios than expected in relation to their fathers' 2D:4D ratio. It was concluded that 2D:4D ratio may be a possible marker for autism which could implicate prenatal testosterone in its aetiology.