Objective:The present study was designed to investigate whether early detection of blood ketone bodies help in diagnosing Diabetic ketoacidosis (DKA) and also to explore whether early changes in blood β-hydroxybutyrate is associated with serum electrolytes and acid-base abnormalities. Research Design and Methods: A total of 122 consecutive type 2 diabetic patients (age 39 ± 15 yrs and body mass index 20.3 ± 2.4 kg/m 2 , mean ± SD) were included in the study. Plasma glucose was measured by glucose oxidase method, glycosylated haemoglobin (HbA 1C ) by high-performance liquid chromatography method, blood β-hydroxybutyrate by biosensor method; urinary acetone was measured by strip based on nitroprusside reaction. Serum urea and creatinine were measured by enzymatic method. Serum electrolytes were measured by ion sensitive electrode technique. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for urinary ketone method were calculated against the blood ketone. Results: The relative frequencies of DKA, using urinary ketone and blood ketone criteria, were 15.6% (19 out of 122) and 13.9% (17 out of 122), respectively. In contrast, 50% (61 out of 122) patients showed hyperketonemia. Using blood ketone as the reference method, the sensitivity of urinary ketone measurement was found to be 32.6% and specificity was 93.7%. PPV and NPV of urinary ketone against blood ketone were 73.68% and 71.84%, respectively. The DKA subjects, diagnosed by blood ketone criteria, showed significant biochemical derangements as compared to Non-DKA subjects [serum urea (P < 0.001), creatinine (P = 0.02), sodium (P < 0.001), potassium (P < 0.001), total carbon dioxide (P = 0.02), and osmolality (P = 0.02)]. Correlation analysis shows that electrolytes, blood gas, and acid-base status have highly significant correlation with blood ketone levels (Na + -r = −0.303, P < 0.001; K + -r = 0.449, P < 0.001; Mg 2+ -r = −0.174, P < 0.05; TCO 2 -r = −0.573, P < 0.001; venous blood pH-r = −0.659, P < 0.001, and osmolality-r = −0.273, P < 0.002). No such correlation was found with plasma glucose except that for serum sodium (r = −0.301, P < 0.001). Conclusions: Measurement of ketonuria by nitroprusside reaction has considerable limitations with an error of about 25-30% in detecting the ketonemic status of type 2 diabetic patients. The present data also demonstrated that severity of hyperketonemia, but not that of hyperglycemia, reflects the underlying biochemical derangements in type 2 diabetic patients better.