The data comes in two flavours, adjusted and all (unadjusted). The first set has been corrected for urban effects and other biases by comparing urban stations to nearby rural stations. The adjusted data are only available for a few countries.
I have post-processed the NCDC data by combining all time series for a given station. At the momen this is done very crudely. For each year, the first series with valid data is chosen. I plan to merge these series more intelligently, using all data and taking into account the possibility of biases.
The quality of the data varies. Most problematic are very old data, before observations were standardised enough (1855, 1905, 1950 are common disconinuities), and very new data from the GTS. Most errors have been removed by the quality control system at NCDC, but inevitably some remain. It always pays to have a glance at the time series. It may also be worth it to check whether the coordinates match with those you expect, e.g., in Google Earth - some stations are mislocated. Please report errors back to me and NCDC.
The database is updated from NCDC into the Climate Explorer once every month.