A recent article in Nature continues to gather attention (it costs $32 to read the article, even though the US taxpayers probably paid for most ofthe research). By analyzing temperature data from the across the continent, including readings from the Antarctic peninsula (warming) and the continental interior (cooling), it concludes that result is a continental warming of 0.12 deg C. per decade between 1957 and 2006.
So why all of the fuss? So far there's been four distinct issues raised.
Its not clear that the code and data that they did the calculation with has been made public. Lots of people are saying that it isn't, and you'd think if it was someone would have shown they could reproduce the results with what's been published, but they haven't. (some of the work to recreate it)
Some of the stations that they used data from have large portions of the data missing. (example)
There's documented cases of the weather observation stations used in the calculation being found buried in snow and dug out by humans. Since snow is a nice insulator, this will have some effect. (example)
The temperature readings used in the calculation were weighted heavily towards the warming peninsula. (example)
The first seems like science 101, while the other three demonstrate the subtlety and complexity of measuring and calculating average temperature. If you're reading articles that cite broad temperature change, make sure you read the fine print and understand how they did their calculation.