We hypothesized that differential handling would affect the responses of sheep to various standardized tests. Indeed, such handling did affect sheep performance in the approach test, both immediately after training and 3 weeks later. Gently handled sheep approached more quickly and sniffed the train- er's hand more often than the restrained and control group sheep. In addition, the gently handled sheep from Source l allowed tactile contact more often than the other groups Thus it appears that unforced handling of sheep results in decreased avoid- ance behaviours, characterized by a greater willingness to approach and con- tact humans. However, this attenuation of avoidance responses did not gen- eralize to other situations such as shearing, halter-restraint, or running through a chute. Changes in behaviour were specific to the test measures that most closely paralleled actual training. Recall that the gentle handling situation was an active interaction between the sheep and the trainer, rather than a passive presentation of human stimuli typical of habituation procedures. This may account for the lack of a general socialization to humans across the different testing situations. Although there was a trend towards diminished responsivity to restraint in the head gate, such forced handling did not reduce evasive responses during subsequent interactions with humans. There were no significant differences in behaviour between sheep that were forcibly handled during restraint and those that received no handling. In the present study the decline of avoidance responses owing to forced restraint and handling was not found to be the per- sistent phenomenon anecdotally reported by other authors (Kilgour, 1987; Grandin, 1989 ). However, it is possible that by extending the restrained han- dling period by either using longer sessions each day or continuing handling over several months, negative responses to humans would eventually cease.