The clinical picture of reflex sympathetic dystrophy (RSD) is characterized by a combination of the triad: autonomic, motor and sensory changes. In this study, the grip strength is measured in 29 upper extremity RSD patients. We used the generalizability theory to assess the extent of the disagreement or differences (errors in measurement) within or between observers and interactions between observer-session and repetition of the measurements. The aims of our study were to determine the different sources of variation in grip strength tests and the smallest detectable differences (SDD) as well as the reliability of upper extremity grip strength tests in RSD patients. The main sources of variation of measurement errors were observer, patient/observer interactions and patient/session/observer interaction and a random source. We found that the generalizability theory is useful for estimating the sources of measurement error. Clinical examinations for muscle strength measurements, as a part of a total clinical examination, for example for a disability payment or worker's compensation in case of RSD patients should be done by more than one observer in more than one session and more than once.