Six hundred horses of different ages, heights and breeds were weighed on a weighbridge and had their weights estimated by two weigh tapes, 1 and 2, by a formula, and by a visual estimate. For the population as a whole, the most accurate method was the formula (mean [sd] 98.6 [10.6] per cent) closely followed by weigh tape 2 (98.1 [8.1] per cent). Tape 1 and the visual estimate were the least accurate (112.0 [9.3] and 88.3 [20.1] per cent respectively). When the population was divided into two height groups, the formula and weigh tape 2 were the most accurate for horses < 15 hh (99.6 [5.2] per cent and 99.0 [5.6] per cent respectively), and weigh tape 1 and the visual estimate were 113.5 (6.5) per cent and 88.4 (16.3) per cent accurate respectively. For horses > or = 15 hh weigh tape 1 was most accurate (103.5 [9.1] per cent) and the formula, tape 2 and the visual estimate were 95.5 (13.1) per cent, 91.8 (9.2) per cent, and 89.3 (22.2) per cent accurate respectively. Overall the formula appeared to be the most accurate estimator of a horse's weight. However, owing to individual variation, it is recommended that the weights of horses < 15 hh are estimated by the formula or weigh tape 2, and that the weights of those > or = 15 hh are estimated with weigh tape 1.