The UN's World Meteorological Organization (WMO) has just announced that this past decade has been the warmest on record in the 160 years for which we've been keeping records. BBC has a very nice summary here. The report also shows that 2009 will be about the fifth warmest year in that record. Now, 160 years is not a terribly long record (speaking as a geologist), but the data are telling. Let's do a little back-of-the-envelope math.
If we measure decadal temperatures as deviations from the overall average, only the past three decades (80's, 90's and the "noughties" [you just have to love Brits]) have been above the average. Granted, there are only 16 decades in the dataset, but let's do a little math. There are precisely 3,360 ways in which you could arrange those three decades over the span of 16. Yet, all three are clustered at the end. Furthermore, while there are 6 different ways in which you could arrange the 3 decades in that cluster (80,90,00; 80,00,90; 90,80, 00; 90,00,80; 00,80,90;00,90,80), we get only one in which the temperature deviation increases consistently (the one which we actually observe). That's a probability of 0.0003 of getting the observed "above average" part of the record, or far less than 1 in a thousand. That's definitely not random.