The purpose was to evaluate the inter-visit, inter-observer and intra-observer variation of quantitative and qualitative tendon examinations in vivo for a cohort of asymptomatic volunteers. Eleven healthy male subjects were recruited. The following tendons were assessed by ultrasonography: Achilles tendon, patellar tendon, triceps tendon, extensor pollicis longus, flexor carpi radialis and supraspinatus. For each tendon a quantitative measurement of tendon size was made at a predefined anatomical location. Two experienced sonologists, blind to one another's findings, evaluated each of the tendons independently. Each tendon was evaluated on two occasions 1 week apart. No difference was found to be attributed to variation in tendon size between visits. Inter-observer variation was a source of error with intra-subject, inter-visit measurements proving more reproducible. There was some significant variation between observers. This variation was more marked with some tendon measures than others. Inter-observer variation for triceps, flexor carpi radialis and supraspinatus was most marked. Minimum detectable change in tendons varied from 13 to 57% depending on the plane of scanning and the tendon being examined. Good reproducibility of quantitative tendon measurements can be achieved within a study using two observers by following a defined scanning protocol. However, it is recommended that the same observer perform serial assessments. The data allow minimal detectable changes in tendon size to be calculated.