Thiopurine dose optimization by thiopurine‐S‐methyltransferase (TPMT) or nudix hydrolase‐15 (NUDT15) significantly reduced early leucopenia in Asia. However, it fails to avoid the late incidence (> 2 months). Although laboratory monitoring of 6‐thioguanine nucleotides (6TGN) to optimize thiopurine dose was suggested in White patients the exact association between leucopenia and 6TGN was controversial in Asian patients. In the present study, we aimed to explore whether DNA‐thioguanine nucleotides (DNA‐TGs) in leukocytes, compared with 6TGN in erythrocytes, can be a better biomarker for late leucopenia. This was a prospective, observational study. Patients with inflammatory bowel disease (IBD) prescribed thiopurine from February 2019 to December 2019 were recruited. Thiopurine dose was optimized by NUDT15 C415T (rs116855232). DNA‐TG and 6TGN levels were determined at the time of late leucopenia or 2 months after the stable dose was obtained. A total of 308 patients were included. Thiopurine induced late leucopenia (white blood cells < 3.5 × 109/L) were observed in 43 patients (14.0%), who had significantly higher DNA‐TG concentration than those without leucopenia (P = 4.1 × 10–9, 423.3 (~ 342.2 to 565.7) vs. 270.5 (~ 188.1 to 394.3) fmol/μg DNA). No difference in 6TGN concentrations between leucopenia and non‐leucopenia was found. With a DNA‐TG threshold of 340.1 fmol/μg DNA, 83.7% of leucopenia cases could be identified. Multivariate analysis showed that DNA‐TG was an independent risk factor for late leucopenia. Quantification of DNA‐TG, rather than 6TGN, can be applied to gauge thiopurine therapy after NUDT15 screening in Chinese patients with IBD.