The stable oxygen isotope ratios of whole tree-ring α-cellulose (δ18OWR) have been interpreted as an indicator of early summer hydroclimate in the Meiyu region of East Asia. However, the underlying physical mechanism often remains unclear. Here we provide a mechanistic understanding through intra-annual tree-ring oxygen isotope analysis and process-based δ18OWR modelling over the period 1979–2006. The selected tree species for analysis is Pinus taiwanensis, whose δ18OWR exhibit the strongest linear relationship with relative humidity (RH) in June. The results indicated that the June RH signal is predominantly contained in tree-ring earlywood rather than latewood. The strong response of δ18OWR to June RH is not due to the legacy effect. Using the proxy system model (PSM) of δ18OWR, we obtained a modeled δ18OWR time series that is significantly positively correlated with the measured δ18OWR time series. The modeled and measured δ18OWR series show similar relationships with monthly RH. Sensitivity experiments with PSM revealed that the June RH signal is originated from the oxygen isotopes of source water and leaf water. Rapid cellulose formation in June plays a role in enhancing the June RH signal. Our study demonstrates how δ18OWR record early summer hydroclimate signals from a process perspective, and that the PSM is effective in modelling the interannual δ18OWR variability in the Meiyu region.