AbstractAn experiment was carried out during 1982 in which the effects of three differing harvesting systems on the field losses occurring during the conservation of grass as silage were examined. The treatments were either (a) harvesting herbage directly by means of a flail harvester (unwilted flail, UF), or (b) pre‐cu ing of herbage with a rotary drum mower and lifting using a precision‐chop forage harvester with wilting (WP) and without wilting (UP). Herbage was harvested on 26 May, 21 July and 7 September 1982. The mean yields of herbage produced from plots harvested by the three systems were not significantly different, with dry ma er (DM) yields of 12470, 12300 and 12230 kg ha‐1for the UF, UP and WP treatments respectively. However, field losses with the UP and WP treatments were greater than with the UF treatment (P<0·01). As a result the yields of herbage ensiled with the UP and WP treatments, 11690 and 11320 kg DM ha‐1respectively, were significantly lower than with the UF treatment, 12710 kg DM ha‐1(P<0·05). Wilting of first cut herbage for a period of 72 h resulted in a significant reduction (P<0·05) in digestible organic ma er (DOM) concentration (determined in vitro), whereas small increases in DOM concentration were observed following wilting of second and third harvest material.It is concluded that, even under favourable weather conditions, both pre‐cutting and wilting of herbage prior to harvesting can result in considerable losses of nutrient