Virtual fencing (VF) is a modern fencing technology using Global Positioning System-enabled collars which emit acoustic signals and, if the animal does not respond, electric pulses. Studies with cattle indicate successful learning and no distinct negative impact on the animals' behaviours and stress level. However, the number of studies testing VF with goats is relatively small. In this study, we used VF collars to test a VF training protocol recently applied to heifers to assess the development of goats' learning to avoid the electric pulse, their behaviour, and faecal cortisol metabolites (FCMs) as an indicator for physiological stress in a grazing experiment. Twenty adult 'Blobe' goats with offspring were divided into two groups and assigned to the VF or physical fencing treatment in a cross-over design with two periods of 12 days each. The VF treatment involved a virtual fence at one side of the paddock, to which the goats were gradually introduced over the first 2 days (additional physical fence or posts as visual support). On day eight, the grazing areas were enlarged by shifting the virtual fence and one side of the physical fencing treatment. The experiment lasted 4 h per day. During this time, the following behaviours were recorded via instantaneous scan sampling of all goats every 2 min: grazing, lying, standing, standing vigilant, walking, and running. Additionally, faecal samples were collected once, or twice daily and FCM concentrations were measured. The VF collars delivered the number of acoustic signals and electric pulses and the duration of the acoustic signals. The daily number of acoustic signals and electric pulses of each goat was used to calculate a 'success ratio'. A significant increase in the success ratio and a general decrease in the signal duration indicate the successful association of acoustic signals and electric pulses at the group level. Behavioural analyses revealed no clear influence of the VF treatment except for standing vigilant. Virtually fenced goats stood significantly more vigilant than physically fenced ones. However, free-moving kids could have had an influence. The VF treatment had no significant effect on the FCM concentrations, which decreased significantly over time. In summary, goats showed signs of learning when avoiding receiving electric pulses by responding appropriately to the acoustic signals. A higher occurrence of vigilance behaviour may suggest insecurity, but FCM concentrations did not indicate increased physiological stress. Future research needs to confirm these results and test VF with goats under practical conditions.