Data sources Medline, Best Evidence, the Cochrane Database of Systematic Reviews, Embase, abstracts of related articles and the reference lists of the retrieved articles were searched by hand.Study selection To be selected, work had to be in the form of controlled studies (minimum-internal control with pre-intervention and post-intervention imaging) of Class II malocclusions treated with the Herbst appliance. Magnetic resonance imaging (MRI) or computerised tomography (CT) scans or tomography (axially or horizontally corrected) had to have been used to image the temporomandibular joint (TMJ). Case reports were excluded.Data extraction and synthesis Eligibility was determined by two authors and agreement reached by discussion. A qualitative synthesis of data was carried out.Results Of 80 studies identified, only five met the inclusion criteria. All studies used internal controls with pre-treatment and post-treatment imaging. Four studies used MRI and one used tomograms. The four MRI studies used overlapping patient samples and were not considered to be independent evidence. The MRI studies did not provide conclusive evidence of osseous remodelling or condyle position change. The tomography study demonstrated minor condyle position change. Methodological deficiencies prevented major conclusions regarding disc position.Conclusions There is a need for randomised controlled studies with serial MRI and tomography to establish short- and long-term effects of Herbst treatment on TMJ osseous and soft tissue morphology. © 2004 EBD.