Goat production systems increasingly involve indoor feeding. To facilitate manger design, there is need for information on the ability to reach for food, such as that for cattle given food through tombstone barriers. In the first study, 20 castrated and 20 non-pregnant female Saanens of mean live weight (M) 39·3 (s.d. 14·42) kg were trained to reach, through a vertical tombstone barrier, for concentrate meal placed on a horizontal platform attached to the barrier. The barrier allowed the neck but not the shoulders to pass through. It was hypothesized that horizontal reach forwards (F, distance from mid point of barrier to uneaten meal) and sideways (S, distances sideways from mid point of barrier to uneaten meal adjacent to barrier) would be a function of height of platform above the floor and size of goat. Because of size, eight goats (mean M, 16·4 kg) were unable to reach meal when the platform height was 75 cm. Mean (s.e.) values for F at platform heights 0, 25, 50 and 75 cm were 55·9 (0·68), 58·8 (0·64), 57·8 (0·69) and 41·1 (1·23) cm respectively. Values for S were smaller, but followed a similar pattern (49·4 (0·64), 52·4 (0·56), 53·2 (0·53), 36·6 (1·17) cm). Linear regression of either F or S on M and linear body dimensions (e.g. body length, neck and head length, neck-joint height, withers height) showed high correlations with R2 values being 0·8 or more. In the second study, 10 castrated and 10 non-pregnant females (M, 48·7 (s.d. 14·56) kg) were trained to reach through the tombstone barrier for concentrate pellets 'glued', using molasses, on to a vertical plate. It was hypothesized that vertical reach (V, distance from floor to uneaten pellets) would be a function of distance between barrier and plate (20, 30, 40, 45, 50 cm), height of step (0,14·2, 28·4, 42·6 cm) on which goats placed their forelegs and size of goat. Four goats (mean M, 29·9 kg) were unable to reach pellets when the plate was at 45 and 50 cm. Mean (s.e.) V values decreased with increasing plate distance and increased with step height (e.g. at 0 cm step, 122·4 (1·96), 116·9 (2·05), 109·4 (2·16), 103·9 (2·49) and 96·5 (2·75) at 20-; 30-; 40-; 45- and 50-cm plate distances respectively; at 20-cm plate distance, 122·4 (1·96), 137·5 (2·00), 151·3 (1·90) and 164·3 (2·00) cm at 0-; 14·2-; 28·4-and 42·6-cm step heights respectively). Linear regression of V on body weight and body dimensions (e.g. body length, heart girth, withers height) showed high correlations (R2 > 0·8). It is concluded that for practical purposes, body weight is a satisfactory predictor of reach. The results support the hypotheses. The data will facilitate the design of mangers for goats with body weights and dimensions in the range of those used in the present study.