Although the conventional in situ ruminal degradability method is a relevant tool to describe the nutritional value of ruminant feeds, its need for rumen-fistulated animals may impose a restriction on its use when considering animal welfare issues and cost. The aim of the present work was to develop a ruminal degradability technique which avoids using surgically prepared animals. The concept was to orally dose a series of porous bags containing the test feeds at different times before slaughter, when the bags would be removed from the rumen for degradation measurement. Bags, smaller than those used in the conventional nylon bag technique, were made from woven nylon fabric, following two shape designs (rectangular flat shape, tetrahedral shape) and were fitted with one of three types of device for preventing their regurgitation. These bags were used in two experiments with individually housed non-pregnant, non-lactating sheep, as host animals for the in situ ruminal incubation of forage substrates. The bags were closed at the top edge by machine stitching and wrapped in tissue paper before oral dosing. Standard times for ruminal incubation of substrates in all of the tests were 4, 8, 16, 24, 48, 72 and 96 h before slaughter. The purpose of the first experiment was to compare the effectiveness of the three anti-regurgitation device designs, constructed from nylon cable ties (‘Z-shaped’, ARD1; ‘double Z-shaped’, ARD2; ‘umbrella-shaped’, ARD3), and to observe whether viable degradation curves could be generated using grass hay as the substrate. In the second experiment, three other substrates (perennial ryegrass, red clover and barley straw) were compared using flat and tetrahedral bags fitted with type ARD1 anti-regurgitation devices. Non-linear mixed-effect regression models were used to fit asymptotic exponential curves of the percentage dry matter loss of the four substrates against time of incubation in the reticulorumen, and the effect of type of anti-regurgitation device and the shape of nylon bag. All three devices were highly successful at preventing regurgitation with 93% to 100% of dosed bags being recovered in the reticulorumen at slaughter. Ruminal degradation data obtained for tested forages were in accordance with those expected from the conventional degradability technique using fistulated animals, with no significant differences in the asymptotic values of degradation curves between bag shape or anti-regurgitation device. The results of this research demonstrate the potential for using a small bag technique with intact sheep to characterise the in situ ruminal degradability of roughages.