Dr. Roger Pielke Jr  on his Blog, April 18th writes:

A new analysis of floods around the world has been called to my attention. The new analysis is contrary to conventional wisdom but consistent with the scientific literature on global trends in peak streamflows. Is it possible that floods are not increasing or even in decline while most people have come to believe the opposite?

Bouziotas et al. presented a paper at the EGU a few weeks ago (PDF) and concluded:

Analysis of trends and of aggregated time series on climatic (30-year) scale does not indicate consistent trends worldwide. Despite common perception, in general, the detected trends are more negative (less intense floods in most recent years) than positive. Similarly, Svensson et al. (2005) and Di Baldassarre et al. (2010) did not find systematical change neither in flood increasing or decreasing numbers nor change in flood magnitudes…