The indirect estimation of intake of grazing animals is usually made from faecal index regression equations. Faecal nitrogen content or total faecal nitrogen are widely used to estimate intake. The most precise estimate of intake is obtained for a specific herbage (e.g. Phalaris) and condition (e.g. green, dry, etc,). Lambourne & Reardon (1963) found that the standard error of an estimate of the intake of a single animal, from a general regression equation, was 23%. Arnold & Dudzinski (1963) showed that the precision of estimates from general equations is improved by multiple regression including terms for faecal output. However, as regression relations are obtained from pen-feeding trials and are applied to field grazing, the following discrepancies axe likely to occur. We may have to extrapolate from low intakes in pens to higher intakes in the field; in addition selective grazing in the field can introduce a serious bias. To overcome the latter, when we are not sure about the exact condition of the forage eaten, as when green and dry herbage are both present, an average regression over two or more conditions has to be used. Naturally the precision of estimate then is much lower, because of the uncertainty as to what the animals are in fact selecting.