Sub-clinical ketosis (SCK) significantly affects post-partum dairy cow performance and welfare. A total of 11,327,959 test-day (TD) records over two years on 1.76 million Holstein cow lactations and 2840 farms were processed to ascertain thresholds for milk acetone (mACE) and β-hydroxybutyrate (mBHB) as indicators of SCK on the basis of a significant milk yield loss at the TD. The set thresholds for mACE and mBHB were 0.10 mmol/L and 0.14 mmol/L, respectively. The prevalence of SCK in the population during the first 60 days in milk (DIM) was estimated based on herd size and milk yields, utilizing one or both of these metabolites surpassing their respective thresholds. Analyzing both mACE and mBHB together revealed a higher occurrence of SCK in small herds (fewer than 100 cows) and a lower occurrence in the two most productive milk categories. The prevalence had an inverse relationship with the daily milk yield at 60 DIM, indicating a surprisingly high frequency of low-productivity herds in the risk classes exceeding 30%. These results suggest that assessing SCK prevalence through the combined evaluation of mACE and mBHB is a more effective approach than using the milk fat to protein ratio, especially when considering different herd sizes and daily milk yield at 60 DIM.