Oxygen isotope measurements from speleothems have been used to infer past hydrological cycles because the isotopic composition of precipitation is observed to decrease with increasing precipitation rates over low latitude coastal and island stations (the “amount effect”). Here we show that local precipitation amount can be inferred from the differences in oxygen isotopic composition if the measurement site is in a subtropical coastal region where water vapor is transported directly from the ocean. The amount effect works because vapor from local evaporation and vapor imported from other regions have distinctly different isotopic compositions – the $\delta^{18}$O of vapor flux from evaporation is ~10‰ higher than the $\delta^{18}$O of imported vapor over subtropical oceanic regions – and precipitation is usually higher over the area with larger convergence. Thus, $\delta^{18}$O in precipitation can differentiate between the contribution of local evaporation and moisture convergence if the isotopic composition of local evaporation and or transported vapor remains relatively constant. Our work indicates that sites further inland may not be suitable for studying the changes in local precipitation amount but still reflect the combined changes of precipitation amount and changes in vapor transport pattern. We have shown that Hulu Cave is a reasonable site to interpret precipitation from $\delta^{18}$O$_p$: estimated precipitation difference between present-day and Last Glacial Maximum (LGM) from measured $\delta^{18}$O$_p$ difference (1~2‰) would be 0.8–2.2 mm/day (lower during the LGM) assuming our model simulates evapotranspiration reasonably well at this site.