This study determined the distribution of drugs to different milk fractions according to their physicochemical properties. Hydrophilic drugs tend to concentrate in skim milk, whereas lipophilic drugs tend to concentrate in cream. The concentration of a drug in casein is related to its degree of binding to milk proteins. Thus, we aimed to determine whether withdrawal time in whole milk differs from that in cream, casein, and skim milk. Amoxicillin and tylosin were selected as proto type hydrophilic and lipophilic drugs, respectively. The study was conducted in vitro and in vivo to determine whether in vitro conditions reflect the distribution of drugs in the different milk fractions in vivo. The in vivo study was conducted using a crossover design on 6 healthy Holstein dairy cattle. First, amoxicillin (i.m., single dose, 14 mg/kg) was administered to cows. Following a 1-wk washout period, tylosin (i.m., single dose, 15 mg/kg) was administered. Concentrations of amoxicillin and tylosin in milk and milk fractions were measured using HPLC-UV. In the in vitro study, 0.04 to 400 mu g/g of amoxicillin and 0.05 to 50 mu g/g of tylosin were spiked to drug-free milk and the concentrations in milk and milk fractions were measured. In addition, the percentage of total protein in milk and milk fractions was determined. Amoxicillin accumulated more in skim milk than in cream and casein, both in vitro (92%) and in vivo (73%, skim milk-to-whole milk ratio). The distribution of tylosin in whole and skim milk was similar to that of amoxicillin in the in vitro study, in contrast to the accumulation of tylosin in cream seen in vivo. However, the accumulation ratio of tylosin in cream was lower than expected. By either method, tylosin was less concentrated in casein than in skim milk and cream. The percentage of total protein was similar in skim milk and whole milk and higher than in cream. Thus, amoxicillin accumulates less in cream and casein, suggesting that these fractions would pose a lower risk to the consumer. Tylosin was still present at the maximum residue limit (50 mu g/kg) 24 h after injection in the casein fraction and 48 h after injection in the cream fraction.