This experiment evaluated different strategies for allocating first-cut grass silages to dry dairy cows that had low body-condition score (BCS) at drying off. A total of 48 moderately yielding Holstein-Friesian cows were used, receiving one of three dietary treatments in the dry period and a single lactation diet based on a flat-rate of concentrates and grass silage ad libitum. Throughout the dry period, one group received a low-digestibility silage (harvested 15 June 1998; LL; metabolisable energy (ME) = 10.3 MJ/kg dry matter (DM)) and a second group received a high-digestibility silage (harvested 9 May 1998, HH; ME = 11.7 MJ/kg DM). A third strategy (LH) offered the low-digestibility silage in the early dry period and the high-digestibility silage in the final 3 weeks before calving. The silages had very different crude protein concentrations (144 and 201 g/kg DM) and intakes were widely divergent (10.1 v. 13.5 kg DM/day) across the dry period. No concentrates were fed during the dry period. Silage quality had a very large effect on liveweight change, with treatment means of 0.32 and 1.75 kg/day for LL and HH, respectively. BCS changes followed a similar pattern, though no cows became over-conditioned and blood metabolites were within normal ranges. Increased silage digestibility in the late dry period led to a substantial increase in milk fat concentration and a smaller increase in milk protein concentration, the latter confined to the first full week of lactation. Depression of milk fat appears related to low blood glucose when dry cows in low body condition are fed at a low level. The LH strategy avoided the tendency for lower milk yields and fat concentration that resulted from feeding the low-digestibility silage until calving. This strategy also avoided the higher calf weights that resulted from feeding the high-digestibility silage in the early dry period.