A large number of investigation techniques are used to establish the relationships between the clinical and biological data which are necessary for physiopathological analysis in the field of developmental disorders. It therefore seemed necessary to develop a quantified grouping system, based on developmental assessments, which could allow closer matching between clinical evaluations and biological numerical data. Two hundred and two subjects presenting developmental disorders (autistic disorder, pervasive developmental disorder not otherwise specified and mental retardation) were examined. For each child, a quantification of autistic behaviour, intellectual impairment, neurological signs and language and communication disorders was performed. A cluster analysis of these quantified data elicited four subgroups according to the scores obtained in these four different areas. We showed the value of this approach by applying it to one of the studies of monoamines routinely examined in childhood autism--dopamine and HVA, its main urinary derivative. Moreover, this method revealed a subgroup within the total population which was independent of nosographic classification and which had a particular clinical and biochemical profile. Other applications could follow, for example in the fields of neurophysiology, cerebral imaging, molecular biology and genetics.