Bruce Hall has an excellent presentation of temperature records in the United States on his weblog.
Among his valuable comments, he writes
“The U.S. analysis showed that the late 1990s were indeed hot and had a greater than normal expected level of statewide monthly records. What it also showed, however, was that the 1930s had a much higher frequency of those records. Finally, it showed a sharp tailing off of such extremes beginning with the new century.
I have completed the review of the high temperature extremes through 2008 and there were no additional statewide month high temperature records. An analysis of the 2005 – 2008 data for minimum temperature records will be started shortly.”
His entire posting is worth reading.