Four treatments (T1 through T4), based on pelleted concentrates varying in as fed crude protein (CP) contents (10.8, 12.2, 13.4, and 15.3%, respectively), but theoretically isocaloric (1.67-1.70 Mcal net energy/kg), were compared for efficacy as pasture supplements. Eighteen multiparous cows began the experiment individually not later than 8 weeks postpartum. Five were assigned to T1 and T3 and four to T2 and T4. Daily concentrate allowances were according to milk production, but 4.5 kg was set as the arbitrary minimum. Nocturnal rotational grazing was in 12 0.5-ha paddocks of little fertilized, heterogeneous, gramineous swards, which fluctuated in quality from adequate to marginal during 11 1/2 months of experimentation. Cows assigned to the four treatments in order produced 21.3 ± 3.0, 22.5 ± 4.4, 20.3 ± 2.5, and 20.4 ± 4.5 kg of milk daily during the 5 days before commencing the experiment, and 14.4 ± 1.2, 16.8 ± 3.9, 15.2 ± 2.4 and 14.2 ± 2.4 kg during the 32 weeks of experimentation. Although treatments did not differ significantly (P>.05) over-all, T1 dropped in daily production by 2.7 kg during the initial 4-week interval and persistency decreased in high-producing cows; T2 also caused a large early decline in production, but superior persistency thereafter. Mean milk fat percentages were 2.78 ± .49, 2.68 ± .44, 2.88 ± .33, and 3.17 ± .37 for 7, through T4. Early depression of milk fat was probably due to high concentrate and inadequate fiber intakes. Concentrate intake ranged from 5.82 kg in T4 to 7.1 0 kg in T2 over 32 weeks; milk/concentrate ratio ranged from 2.33 in T3 to 2.45 in T4. Only in T1 was appreciable liveweight (20 kg) temporarily lost early in the experiment. Concentrates containing 13 to 14% CP can be tentatively recommended as supplements to pastures of adequate quality, when fed to meet energy requirements of cows producing up to at least 20 kg of milk daily.