International translation and psychometric testing of generic health outcome measures is increasingly in demand. Following the methodology developed by the International Quality of Life Assessment group (IQOLA) we report the German work with the SF-36 Health Survey. The form was translated using a forward-backward method with accompanying translation quality ratings and pilot tested in terms of translation clarity and applicability. Psychometric evaluation included Thurstone's test of ordinality and equidistance of response choices in 48 subjects as well as testing of reliability, validity, responsiveness and discriminative power of the form in crossectional studies of two samples of healthy persons and longitudinal studies of two samples of pain patients totalling 940 respondents. Quality ratings of translations were favorable, suggesting a high quality of both forward and backward translations. In the pilot study, the form was well understood and easily administered, suggesting high clarity and applicability. Thurstone's test revealed ordinality (in over 90% of the cases) and rough equidistance of response choices also as compared to the American original. On item and scale level, missing data were low and descriptive statistics indicated acceptable distribution characteristics. In all samples studied, discriminative item validity was high (over 90% scaling successes) and Cronbach's alpha reliabilities were above the 0.70 criterion with exception of one scale. Furthermore convergent validity, responsiveness to treatment and discriminative power in distinguishing between healthy and ill respondents was present. The preliminary results suggest that the SF-36 Health Survey in its German form may be a valuable tool in epidemiological and clinical studies. However further work as concerns responsiveness and population based norms is necessary.