Background: As part of the pervasive developmental disorders (PDD), there is a subgroup of individuals reported to have a different onset of symptom appearance consisting of an apparently normal early development, followed by a loss of verbal and/or non-verbal skills prior to 2 years of age. This study aims at comparing the symptomatology of children who displayed a regression and often an associated intellectual disability through investigation of two types of loss, namely language and other skill regression. Methods: This study examined the occurrence of regression in 135 children with PDD, mean age 6.3 years. The sample was composed of 80 (59.4%) children diagnosed with autism, 44 (32.6%) with pervasive developmental disorder-not otherwise specified (PDD-NOS) and 11 (8%) with Asperger syndrome. The Autism Diagnostic Interview Revised (ADI-R) was used to evaluate the type of loss and to characterise associated factors including birth rank, gender and thimerosal exposure through vaccination. Results: A total of 30 (22%) subjects regressed: nine (30%) underwent language regression alone, 17 (57%) lost a skill other than language and four (13%) lost both language and another skill. Significantly higher levels of regression were found in autism (30%) compared with PDD-NOS (14%) and Asperger syndrome (0%). Children who regressed in language skills spoke at a significantly earlier age (X = 12 months) than those who did not regress in this domain (X = 26 months). Parents and interviewers consistently reported developmental abnormalities prior to the loss. ADI-R domain mean scores indicated a more severe autistic symptomatology profile in children who regressed compared with those who did not, especially in the repetitive behaviour domain. Regression was not associated to thimerosal exposure, indirectly estimated by year of birth. Conclusions: A loss of skill, present in one out of five children with PDD, is associated with a slightly more severe symptomatology as measured by the ADI-R, particularly in the repetitive behaviours domain. Furthermore, although abnormalities are often noticed by the caregivers at the time of regression, the ADI-R reveals that other atypical behaviours were in fact present prior to the onset of regression in most cases. None of the secondary factors investigated were associated with regression. In children unexposed to thimerosal-containing vaccines, the rate of regression was similar to that reported in studies of samples exposed to thimerosal, suggesting that thimerosal has no specific association with regressive autism.