The reliability and validity of the eating behaviour in the presence of man (Feeding test) as an index of fear were assessed in farmed blue (Alopex lagopus) and silver foxes (Vulpes vulpes). Repeatability of the Feeding test was good in both species. No further habituation occurred after the fourth successive test in either species. In addition, the behaviour of both species was independent of the person who performed the test. The normal feeding interval, i.e., 24 h, between feed deliveries, was long enough to provide reliable results. The presence of a cage mate did not influence the blue foxes' response in the Feeding test. A significant relationship between the results of the Feeding test and the Tit-bit test in both species and between the Feeding test and fearfulness score in silver foxes indicate that all these tests measure similar features, most probably foxes' fear of humans. Those silver foxes that did not eat in the Feeding test had higher base levels of cortisol than the animals that did eat, providing further support for the above conclusion. The present study demonstrates that the Feeding test is a reliable, i.e., repeatable and free of random errors, and fairly valid fear test for blue and silver foxes. The Feeding test seems likely to give good results in measuring fear in farmed blue and silver foxes, but further investigations will be needed to fully validate it, especially for blue foxes.