Abstract 32 tritium analyses were performed on 13 different water sources within Soreq Cave, Israel; eight of the water sources were analyzed once each, and the other five between 3 and 6 times each, on samples collected between 1990 and 2000. The results were compared with similar analyses of a 1981–82 collection and with the tritium concentrations of the annual rainfall from 1952 to 1998. The tritium in several of the cave waters was so high (up to 110 TU) that they must have originated mainly in the peak tritium rain year 1964. Most of the other cave waters had tritium concentrations, which though somewhat lower, must have originated largely from rain that fell between 1962 and 1966. The consequences of these tritium observations are that: (1) it generally takes 26–36 years for rainwater to percolate from ground surface to Soreq Cave ceiling; and (2) the type of flow undergone during this percolation is in several cases very similar to ‘piston flow’, hence the annual packets of rainfall reach the cave without undergoing much mixing with either earlier or later rains. In about one-third of the cases, tritium was too low to determine the percolation time. Finally, the time variations of the tritium concentrations in specific cave water sources seem to indicate that the conduits which bring water to these sources may have undergone physical changes which affect the amount of time it took for rain to reach the cave ceiling.