Eighty-nine autumn-calving first calf and adult Friesian cows participated in an experiment on the effect of feeding over three lactations on milk production and live weight change. Fixed daily allowances of digestible energy (DE) formed two of the treatments (high, H; moderate, M). Diets of similar composition were used for both treatments and rations were weighed daily for each cow. The cows within these treatments were re-randomized to H or M at second and again at third parturition on experiment. A further treatment (ALF), applied continuously over three lactations, consisted of the M allowance of compound feed, weighed daily for each cow, plus ad lib. weighed, group-fed forages. The ALF animals were randomized for each lactation into two groups both of which received the same total compound feed allowance over the first 26 weeks of lactation. For one group (Flat) equal amounts were given daily whilst for the other group (Step) the daily amount was decreased monthly. After week 26 equal rations were fed. Hay, maize silage and grass silage formed the forages in winter. Grass, cut for the H and M groups but grazed for the ALF group, provided the summer forage. Energy intakes covered some 80-110% of requirements (Alderman et al. 1975) Yields of milk and of milk solids responded similarly for both parities. In the first experimental lactation, treatment H led to greater yields compared with M. H also led to smaller losses of live weight in early lactation, equal gains in mid lactation, and smaller gains in late lactation and the dry period, compared with M. Extension of H into a second lactation increased the advantage in milk and solids yields observed in the first lactation on experiment. Recovery of body reserves on treatment M continued. Treatment H in a second lactation on experiment after M in the first lactation led to even greater compensatory gains in live weight at the expense of milk production. There was no effect in the third lactation on experiment of treatments applied in the first lactation. Treatments H and M applied factorially over lactations 2 and 3 gave the same pattern of treatment effects as in lactations 1 and 2. Treatment ALF broadly supported the same milk yield and live weight change as treatment H but improved fat, protein and lactose yields. Within treatment ALF, Flat and Step distribution of compound led to equal performance. Multiple lactation effects of ALF equalled those of H. The effects on milk composition of H compared with M treatment were variable. In general an advantage accrued to ALF over M but without long term effects.The effect of variation in intake on performance of the dairy cow has been extensively documented for short periods within lactations (Broster, 1972), and more so for milk production than live weight change. However, the evidence on the size and development of effects of variation in feeding over protracted periods within the adult life span of the dairy cow, e.g. some four lactations in the UK, is extremely limited , both for plane of nutrition and for d...