Despite 20 years of intensive effort to understand the global carbon cycle, the budget for carbon dioxide in the atmosphere is unbalanced. To explain why atmospheric CO2 is not increasing as rapidly as it should be, various workers have suggested that land vegetation acts as a sink for carbon dioxide. Here, I examine various possibilities and find that the evidence for a sink of sufficient magnitude on land is poor. Moreover, it is unlikely that the land vegetation will act as a sink in the postulated warmer global climates of the future. In response to rapid human population growth, destruction of natural ecosystems in the tropics remains a large net source of CO2 for the atmosphere, which is only partially compensated by the potential for carbon storage in temperate and boreal regions. Direct and inadvertent human effects on land vegetation might increase the magnitude of regional CO2 storage on land, but they are unlikely to play a significant role in moderating the potential rate of greenhouse warming in the future.