Here is the scary CO2 graph; red points are the famous Mauna Loa data, blue points are from the number 1 ice-core study turned up by Google (http://cdiac.ornl.gov/trends/co2/lawdome.html)
Here is the HadCRUT3 global temperature dataset for the same period at annual resolution; the solid black line is the best estimate value, the red band gives the 95% uncertainty range caused by station, sampling, and measurement errors, the green band adds the 95% uncertainty range due to limited coverage, and the blue band the 95% bias range due to bias errors.
Bias errors are urbanisation (the 'heat island' effect) and changes in thermometer exposure protocols over time. They have taken an urbanisation bias value of 0.055 C per century for their land values, which seems to be a consensus value. Some researchers they cite (i.e., not fruit loops) claim this may be as high as 0.3 C per century, but this figure incorporates both land and sea data so is not likely to be out by as much as that.
Here is my quick and dirty correlation of the two graphs above, with one data point every five years:
My naive extrapolation of graph (1) and graph (2) suggests that the mean global temperature ought to be bopping up and down one side or another of 0.0-0.2 degrees above today's mean c. 2030.