Abstract. Although radar-based quantitative precipitation estimation (QPE) has been widely investigated from various perspectives, very few studies have been devoted into extreme rainfall QPE. In this study, the performance of KDP-based QPE during the record-breaking Zhengzhou rainfall event occurred on 20 July 2021 was assessed. Firstly, OTT disdrometer observations were used as input to T-matrix simulation and different assumptions were made to construct R(KDP) estimators. Then, KDP estimates from three algorithms were compared for obtaining best KDP estimates, and gauge observations were used to evaluate R(KDP) estimates. Our results in general agree with previous known-truth tests, and provide more practical insights from the perspective of QPE applications. For rainfall rates below 100 mm h-1, R(KDP) agrees rather well with gauge observations, and the selection of KDP estimation method or controlling factor has minimal impact on QPE performance provided that the used controlling factor is not too extreme. For higher rain rates, significant underestimation was found for R(KDP), and a smaller window length results in higher KDP thus less underestimation of rain rates. We show that the “best KDP estimate”-based QPE cannot reproduce the gauge measurement of 201.9 mm h-1 with commonly used assumptions for R(KDP), and potential responsible factors were discussed. We further show that the gauge with the 201.9 mm h-1 report was located at the vicinity of local rainfall hot spots during 16:00 ∼ 17:00 LST, while the 3-h rainfall accumulation center was located at the southwest of Zhengzhou city.