Jersey cows were used in a change-over design to examine the effect of three levels of energy intake (lucerne hay at 100, 75 and 50% ad lib.) on milk yield, milk composition, fatty acid composition of milk fat, and blood metabolites (non-esterified fatty acids, glucose and total ketones) to determine which measurement was the most accurate indicator of intake of energy. Milk yields averaged 9.9, 8.7 and 7.2 kg/cow/day with relative intakes of 100, 75 and 50% of ad lib. When energy was restricted the proportion of C4–C16 fatty acids in milk fat decreased (72, 69 and 59%), while the proportion of oleic acid increased (15, 18 and 26%). These changes occurred within approximately 6 days on new energy levels. Fore milk and strippings had similar fatty acid proportions. Restriction of energy reduced the solids not fat, protein and casein contents of milk, and increased its butter fat percentage. Non-esterifred fatty acid levels in blood plasma increased with restriction of feed (348, 528 and 579 µ-equiv./l). Glucose and ketone bodies of blood averaged 58 mg/100 ml and 9.1 mg/100 mi respectively, and did not vary between treatments. It is concluded that milk production is the most sensitive indicator of the intake of digestible energy where change-over designs are used. However, when individual animal variation is not removed in the analysis, the intake of energy is most closely correlated with the fatty acid composition of milk fat (r = 0.73 and –0.74 for C4–C16 acids and oleic acid respectively). Significant correlations with the protein to fat and casein to fat ratios of milk were also measured (r = 0.64 and 0.63 respectively). There was a poor relationship between energy intake and blood composition (r = –0.25 for non-esterified fatty acid content).