The diversity of bacteria in the world is very poorly known. Usually less than one percent of the bacteria from natural communities can be grown in the laboratory. This has caused us to underestimate bacterial diversity and biased our view of bacterial communities. The tools are now available to estimate the number of bacterial species in a community and to estimate the difference between communities. Using what data are available, I have estimated that thirty grams of forest soil contains over half a million species. The species difference between related communities suggests that the number of species of bacteria may be more than a thousand million. I suppose that the explanation for such a large number of bacterial species is simply that speciation in bacteria is easy and extinction difficult, giving a rate of speciation higher than the rate of extinction, leading to an ever increasing number of species over time. The idea that speciation is easy is justified from the results of recent experimental work in bacterial evolution.