AbstractThe ability to mix a quality feed is often equated to the quality of the mixer; the ability to weigh ingredients correctly has received little attention. To assess how accurately feed mills weigh their ingredients, 14 feed mills specialising in swine diets were surveyed, which yielded 8432 data points (for 229 ingredients and 11–44 batch records per ingredient within mills). Amounts actually weighed (according to scale readings) were compared to calls, and differences were analysed statistically. Feed mills overdosed ingredients by 1.5 ± 16.3%: between mills, overdosing ranged from −0.7 to 13.0%. Within ingredients, weighing variation ranged from 0.6 to 11.1% between mills and averaged 5.2%. Some of the weighing problems observed were attributed to discrepancies between the call size and the scale resolution. For example, weighing 11.3 units (pounds) on a scale with a 2 unit resolution leads to a minimum error of 6%. Such problems occurred for 8.7% of the calls and resulted in a minimum error ranging from 0.01 to 20%, averaging 1.95%. Poor choice of scales was the major source of errors in weighing, and the relationship weighing variation = 10∧ [1.56 − 0.50 log (call/scale resolution)] explained 40% of the variation observed (p < 0.05). Weighing ingredients in the right scale would thus not only benefit feed quality but also reduce diet cost, as it would reduce the overdosing of ingredients.© 2001 Society of Chemical Industry