Autistic regression seems to occur in about a quarter of children with autism. Its cause is unknown. Late-onset autistic regression, that is, after 2 years of age, shares some features with catatonic regression. A working hypothesis is developed that some children with autistic regression suffer from early-onset catatonic regression. This hypothesis cannot be answered from current data and is difficult to address in clinical studies in the absence of definite markers of autistic and catatonic regression. Treatment implications are theoretical and involve the potential use of anticatatonic treatments for autistic regression. Focus is on electroconvulsive therapy (ECT)--an established but controversial treatment that is viewed by many, but not all, as the most effective treatment for severe, life-threatening catatonic regression. Clinical trials of ECT in early- or late-onset autistic regression in children have not been done yet. The effects of electroconvulsive seizures--the experimental analogue of ECT--should also be tested in gamma-aminobutyric acid-ergic animal models of autistic regression, autism, catatonia, and other neurodevelopmental disorders. Purkinje cell survival and neurogenesis are putative outcome measures in these models.