Significant warming in extreme temperatures during the last few decades has been detected at annual, seasonal and daily timescales across a range of spatial scales. We present results from formal statistical analyses based on optimal fingerpriting which show that the warming is primarily driven by human influences on the climate. Moving to sub-continental regions, internal variabilty tends to dominate over the forced climatic response. We developed a new methodology that uses constraints from a global attribution analysis to construct regional temperature distributions with and without the effect of anthropogenic forcings. We show that human influences have at least quadrupled the likelihood of having a year warmer than the warmest year in the observational record in 23 out of the 24 regions examined. Finally, we discuss the development of a new system in the Hadley Centre for the attribution of weather and climate extreme events. The system uses very large ensembles of atmosphere-only models runs with prescribed sea surface temperatures and estimates the changing likelihood of events under the influence of anthropogenic forcings. First applications of the new system to high-profile events in year 2010, like the July heatwave in Moscow, will be presented.