The objective of this study was to evaluate the characteristics of five ketone tests for use with milk to detect subclinical ketosis [defined as >1200 µmol of β-hydroxybutyrate (BHBA)/L of blood] in dairy cows. The tests studied were the Ketolac® BHB strip to detect BHBA (Hoechst, Unterschleiβheim, Germany) and four tests based on sodium nitroprusside to detect acetoacetate (Rothera tests) [Ketostix® strip (Bayer, Etobicoke, ON, Canada), Bioketone powder (Societé d’ Analyses Biopharmaceutiques, Laval, QC, Canada), Ketocheck TM powder (Great States, St. Joseph, MO), and Utrecht powder (University of Utrecht, Utrecht, The Netherlands)]. Milk samples (n = 529) from 266 cows in 25 Ontario dairy herds were used for this investigation. The Ketolac® BHB strip at 50 and 100 µmol of BHBA/L of milk showed sensitivities of 92 and 72%, respectively. This test was more sensitive for subclinical ketosis than were any of the Rothera tests. The sensitivities of the Rothera tests were 43, 33, 28, and 5%, respectively, for Utrecht powder, Bioketone powder, Ketocheck TM powder, and Ketostix® strip. The Rothera tests were highly specific, as was the Ketolac BHB strip at 200, 500, and 1000 µmol of BHBA/L of milk (specificity, >97%). The prevalence of subclinical ketosis was highest during the first 6 wk of lactation, and a peak was detected during the 2nd wk of lactation. Of the tests evaluated in this study, Ketolac® BHB strip was the most useful to monitor subclinical ketosis in dairy herds.