The prevalence, as well as the effect of subclinical ketosis on daily milk yield, was observed using 1.299,630 test-day records collected from January 2000 to December 2005 on 73,255 Slovenian Holstein cows. Subclinical ketosis was indicated by the fat to protein ratio (F/P ratio) higher than 1.5 in cows that yielded between 33 to 50 kg of milk per day (Eicher, 2004). The ketosis index was defined in relation to the timing of subclinical ketosis detection to the subsequent measures of test-day milk yields. The effect of subclinical ketosis on test day milk yields were studied separately for each parity using the mixed model analysis. The statistical model included the fixed effect of ketosis index, calving year-month, lactation stage and random effect of animals included in the study. The prevalence peak of subclinical ketosis occurred in the first month of lactation. Subclinical ketosis had a significant negative effect on daily milk yield. Decrease in milk yield in the amount of 4.21 kg/day; 2.73 kg/day; 2.78 kg/day; 2.83 kg/day; and 3.72 kg/day in each parity were determined within 35 days after the detection of subclinical ketosis. The decrease continued in subsequent milk controls. The research results show that test-day records could be a useful tool for early detection of subclinical ketosis.