Vitamin D in milk is determined by a slight modification of the method of Sliva et al. [J. AOAC Int. (1992) 75, 566-571] for infant formula and enteral nutritional products. The sample is saponified for 30 min at 60 degrees C and extracted overnight into 60 mL of hexane. The hexane layer is washed, neutralized, and taken to dryness with a rotary evaporator. The sample is reconstituted in hexane and applied to 500 mg of Florisil in a solid-phase extraction column. Vitamin D is eluted with isopropyl alcohol. The eluate is evaporated to dryness under N2, and the sample is reconstituted in 1.0 mL of acetonitrile. The extract is analyzed on a C18 liquid chromatographic column (250 x 4.6 mm, 5 microns particle size) with UV detection at 265 nm. Milk samples of various fat content (i.e., skim, low fat, and whole milk) were analyzed. Spiked recoveries gave means of 81-96%; recoveries were inversely related to fat content. Assay precision ranged from 3.2 to 8.6%. The method can measure vitamins D2 and D3 individually, and no difference in the recoveries of the 2 vitamins was observed. Thus, vitamin D2 can be used as an internal standard for quantitating vitamin D3, and vice versa. The method is satisfactory for use in screening of milk for vitamin D content.