Quantifying rainfall recharge thresholds, including their spatial and temporal heterogeneity, is of fundamental importance to better understand recharge processes and improving estimation of recharge rates. Caves provide a unique observatory into the percolation of water from the surface to the water table at the timescale of individual rainfall recharge events. Here, we monitor nine infiltration sites over six years at a montane cave site in south eastern Australia. Six of the drip hydrology time series have up to ~100 hydrograph responses to rainfall over the monitoring period, three sites do not respond to rainfall events. We use two approaches to quantify rainfall recharge thresholds. At an annual timescale, for all nine drip sites, the total annual percolation water volume was determined for each year of data. Daily rainfall recharge thresholds were then determined by maximising the correlation of annual percolation water volume and total precipitation above a variable daily threshold value. The annual recharge amount methodology produced rainfall recharge thresholds for seven sites, where high and significant correlations (rank correlations > 0.75) occur for daily precipitation thresholds between 6 mm and 38 mm/day. No rainfall recharge thresholds could be obtained from one site which had a low and constant annual drip amount, and from one site which exhibited ‘underflow’ behaviour. At an event timescale, for the six sites which had a hydrograph response to rainfall, the 7-day antecedent rainfall amounts were determined. Minimum 7-day precipitation amounts prior to a hydrograph response for specific drip sites were in the range 13–28 mm and 75% of all recharge events had a 7-day antecedent precipitation between 20.7 and 38.1 mm. Combining all drip water monitoring sites and analysing the data by month identifies a seasonal variability in the minimum 7-day antecedent precipitation necessary to generate potential recharge, from 15 to 25 mm in winter to >50 mm in February and March. We apply a simple water budget model, driven by P and ET and optimised to the observed potential recharge events, to infer a ‘whole cave’ soil and epikarst storage capacity. This storage capacity is between ~50 mm (using potential evapotranspiration, 92% of events simulated successfully) to ~60 mm (using actual evapotranspiration, 79% of events simulated successfully). Modelling of individual drip sites identifies spatial heterogeneity in soil and epikarst storage capacities. Our approach using multiple methodologies allows the comparison between both daily and weekly rainfall recharge thresholds and modelled soil and epikarst storage for the first time.