Adult-onset primary torsion dystonia (AOPTD) has an autosomal dominant pattern of inheritance with markedly reduced penetrance; the genetic causes of most forms of AOPTD remain unknown. Endophenotypes, markers of sub-clinical gene carriage, may be of use detecting non-manifesting gene carriers in relatives of AOPTD patients. The aim of this study was to compare the utility of the spatial discrimination threshold (SDT) and temporal discrimination threshold (TDT) as potential endophenotypes in AOPTD. Data on other published candidate endophenotypes are also considered. Both SDT and TDT testing were performed in 24 AOPTD patients and 34 of their unaffected first degree relatives; results were compared with normal values from a control population. Of the 24 AOPTD patients 5 (21%) had abnormal SDTs and 20 (83%) had abnormal TDTs. Of the 34 first degree relatives 17 (50%) had abnormal SDTs and 14 (41%) had abnormal TDTs. Discordant results on SDT and TDT testing were found in 16 (67%) AOPTD patients and 21 (62%) first degree relatives. TDT testing has superior sensitivity compared to SDT testing in AOPTD patients; although false positive TDTs are recognised, the specificity of TDT testing in unaffected relatives is not determinable. The high level of discordance between the two tests probably relates methodological difficulties with SDT testing. The SDT is an unreliable AOPTD endophenotype; TDT testing fulfils criteria for a reliable endophenotype with a high sensitivity.