Ketosis in dairy cattle is primarily diagnosed based on the concentrations of ketone bodies in the blood, milk, or urine. Cow-side tests using these fluids are available for rapid detection of elevated concentrations of ketone bodies. Although these tests have been extensively validated, the performance of different tests has not been compared over time. Our objectives were to investigate the relationship between point-of-care diagnostic tests measuring the concentrations of β-hydroxybutyrate (BHB) in blood (BT; Precision Xtra, Abbott Laboratories), BHB in milk (MT; Keto-Test, Elanco), and acetoacetate (AcAc) in urine (UT; Ketostix, Bayer Corporation) through cases of ketosis. Holstein cows (n = 148) were screened daily for hyperketonemia (HYK; blood BHB ≥1.2 mmol/L) from 3 to 16 d in milk (DIM); moreover, milk and urine samples were collected concomitantly and tested for ketones (ketosis thresholds: 100 µmol/L milk BHB and 5 mg/dL urine AcAc). Of the animals screened (n = 148), 74% were diagnosed with HYK. When diagnosed with HYK, cows were treated with propylene glycol orally once daily for 5 d. After the day of diagnosis (d 0), hyperketonemic cows were retested with BT, MT, and UT for 3 d (d 1, 2, and 3). We assessed the diagnostic test performance and time to ketosis (survival analyses and Cox proportional hazards models) of MT and UT compared with BT. Considering all paired samples (before and after diagnosis of HYK), MT had 61% sensitivity and 91% specificity, whereas the UT had 77% sensitivity and 94% specificity compared with BT. The specificity of MT and UT increased from d 0 to d 1, decreased on d 2, and increased on d 3. The median time to diagnosis of ketosis in blood was 5 DIM (95% CI 5 to 7 DIM); moreover, MT and UT had 2 d greater median time to diagnosis of ketosis compared with the BT [7 DIM (6 to 11 d); and 7 DIM (6 to 13 d), respectively]. We concluded that the UT is a more sensitive predictor of blood BHB concentration than the MT. The UT and MT tests diagnosed ketotic cows approximately 2 d later than the BT. The possible consequences of delay in detection of ketosis in milk and urine should be investigated.