The objective of the present study was to examine the effect of a Linear Programming (LP) model developed to concomitantly optimize cropping plans and animal diets to minimize whole dairy farm feed costs. The optimizations were carried out on data from 15 Italian dairy farms. All herds were high yielding Holstein-Friesian cows (37.5 ± 2.3 kg of ECM), housed in freestall barns, fed TMRs, and without access to pasture. Nutritional requirements were kept equal to the actual farm practice. These included DMI, RDP, RUP, NEL, NDF, ADF, f-NDF, which were group calculated according to NRC (2001) equations. Production levels were considered to remain constant as the nutritional requirement remained unchanged. Other farm characteristics such as herd structure, nutritional grouping strategies, feed consumption, cropping plan, intrinsic farm limitations (e.g., silage and hay storage availability, water for irrigation, manure storage) and on-farm-produced forage costs of production were collected from each farm for year 2017. The optimized scenario resulted in different diets and cropping plans with different feed allocation. The total feed cost (€/100 kg of milk) was greater in the actual (€20.4 ± 2.3) than the optimized scenario (€19.0 ± 1.9), resulting in a 6.7% feed cost reduction with a range between 0.49% and 21.6%. This means €109 ± 96.9 greater net return per cow per year. The implementation of our proposed LP decision support tool for the optimal allocation of the nutritional resource and crops in a dairy herd has the potential to reduce feed cost of diets and improve the farm feed self-sufficiency. Table 1 (Abstr. 100). Differences in feed costs, energy, and protein feed self-sufficiency among farms between optimized scenario minus actual situation, expressed as % of herd requirement (mean ± SD)

Variable	Difference (%)	P-value
Feed costs (€/lactating cow per day)	−6.65 ± 4.32	0.057
Feed self-sufficiency (energy)	18.25 ± 20.59	<0.05
Feed self-sufficiency (protein)	10.34 ± 9.29	<0.05