An experiment was designed to test whether responses to variation in plane of nutrition conformed to a linear model or to a diminishing response curve model and to examine the influence of type of energy-yielding nutrient used in the ration on the response. Lactating Friesian cows (n = 18; mean, 126 DIM) were used in a Latin square experiment with three 4-wk periods. Diets consisted of hay and concentrates 40:60 (wt/wt, DM basis). The concentrates were based on grain, sugar beet pulp, or an equal mixture of the two. Each cow was offered low, medium, or high amounts of feed within the Latin squares, and feed type was compared between squares. The high amount was sufficient to maintain current body state and predicted final milk yield. Medium and low amounts were set to 1.5 and 3.0 kg of DM/d lower than the high amount, respectively. The source of dietary energy did not affect performance or response to amount of feed. The amount of feed had a highly significant linear effect on milk yield. The time taken for responses to a change in amount of feed to stabilize was 1 wk; milk yield did not reach a plateau but declined at a constant rate for the remaining 3 wk of the period. The rate of decline was significantly affected by amount of feed.