A recent study in the US by the American Chemical Society is getting a lot of press because it found that around 90% of banknotes they tested had traces of cocaine. This was up from 67% a couple of years ago. The survey also included some international locations and the figures for Tokyo and Kyoto were 20% and 10% respectively.
The New Scientist has now given some reasons why we need to treat these numbers with caution. Apart from questioning the survey methodology, they point out that clean bills can be contaminated when they come into contact with a note that has been used for snorting coke. That makes the Tokyo number even more interesting because a note that has been rolled up would not often come into contact with new notes so, arguably, there might be less of a contamination effect in Japan.
One report suggests, however, that they tested less than 20 bills in Japan. I'm not sure what sample size would be better but that seems pretty low to me.