They make interesting reading.
The above figures represent the flows sampled over 16-hour periods across Chelsea, Albert, Battersea, Putney and Wandsworth bridges.
You'll note that the number of cycles was stuck in the 3% - 4% range until 2002, at which point the numbers have turned sharply up. The 2010 figure is calculated on a different basis to other years, because Albert Bridge has been closed and the average is taken over four bridges rather than five. However, it is not clear what effect the bridge closure may have had.
Motor traffic has been declining since 2000 both in absolute terms and as a percentage of modal share. We know that recessions are good for cycling, so it's likely that the last couple of years' increase in modal share will have been partly due to economic conditions. The real test is whether the trend continues when the economy is growing, but that may not happen soon.
You could of course spin this data in many different ways. TfL might claim that this is evidence that its approach to cycling is working. Others might claim that there is clearly substantial underlying demand, and the trivial amounts spent on cycling don't match its increasing importance; under-investment means much demand is being suppressed by the poor cycling infrastructure in London.
One curiousity is why the modal share on bridges is significantly higher than the modal share in the surrounding boroughs. This needs to be explained. There's no readily apparent reason why bridges should attract more cyclists than other roads. Cycling on bridges is easy to measure, so it's possible that cycle journeys are being under-estimated due to the dispersal of cyclists into minor roads which aren't subject to counts. Cyclists can avoid major routes where there are cycle counts, but they cannot avoid bridges if their journey crosses the river.