Tritium concentration was measured in snow deposited at the GRIP site (central Greenland) and at the Vostok station (east Antarctica) from snow pits covering the period 1980–1990. The objective of the study was to investigate tritium concentrations in polar regions several decades after the bomb peak of the sixties and to put them in the context of available data for environmental tritium in the Arctic and the Antarctic over the last five decades. The tritium content of the samples was measured by mass spectrometry using the helium-3 regrowth method. In Antarctica, the tritium concentrations are in the range 70–110 TU. The comparison of the bomb tritium history at different locations show that tritium levels increase moving inland, where vapour pressure becomes extremely low and therefore more sensitive to the intrusion of stratospheric air masses highly enriched in tritium. Although most tritium fallout occurred in the Northern hemisphere, the tritium levels in central Greenland in the 80's, in the range 10–40 TU, are significantly lower than at Vostok. Unlike Antarctica, no such continental effect is observed in Greenland, due to the higher water vapour content of the air masses, as evidenced by the much higher snow accumulation rate. Whereas tritium fallout in Antarctica appears to occur as a result of direct injections of stratospheric tritium during winter, Arctic fallout are the result of the dominant spring injection of stratospheric air at mid-latitude, in line with the deposition of other stratospheric tracers.