This study investigated dexamethasone (DXM) residues in the milk from intramuscularly dosed dairy cows and established the withdrawal time (WT) of DXM in milk. Eighteen healthy Holstein cows were injected with 20 (DXM-1) or 40 mL (DXM-2) of a drug containing 1 mg/mL of DXM. After administering DXM, milk samples were collected from all cows at 12-hour intervals for five days. The DXM residue concentrations in milk were determined by liquid chromatography-tandem mass spectrometry. The correlation coefficient of the calibration curve was 0.9966, and the limits of detection and quantification (LOQ) were 0.03 and 0.1 μg/kg, respectively. The recoveries were 97.0% to 104.0%, and the coefficient of variations was less than 7.22%. After treatment, DXM in DXM-1 was detected above the LOQ in two milk samples at 36 hours and below the LOQ in all milk samples of DXM-2 at 48 hours. Using the WT calculation program WT 1.4, the withdrawal periods of DXM-1 and DXM-2 in milk were established to be two days. In conclusion, the developed analytical method is sensitive and reliable for detecting DXM in milk. The estimated WT of DXM in bovine milk is shorter than the current milk WT recommendation of three days for DXM in lactating dairy cows.