Climate variability and global change studies are increasingly focused on understanding and predicting regional changes of daily weather statistics. Assessing the evidence for such variations over the last hundred years and evaluating the quality of models making predictions for the next hundred requires a subdaily (as opposed to monthly or longer-term average) tropospheric circulation dataset. The only dataset available for the early 20th century consists of error-ridden hand-drawn analyses of the mean sea level pressure field over the Northern Hemisphere. Modern data assimilation systems have the potential to improve upon these maps, but prior to 1948, few digitized upper-air sounding observations are available for such a “reanalysis.” We have investigated the possibility that the quantity of newly recovered surface pressure observations is sufficient to generate useful reanalyses of at least the lower tropospheric circulation back to 1900. Surprisingly, we have found that with an Ensemble Kalman Filter that blends an ensemble of 6-hour numerical weather prediction model forecasts with the available observations, one can produce high-quality reanalyses of even the upper troposphere using only surface pressure observations. For the beginning of the 20th century, the errors of such upper-air circulation maps over the Northern Hemisphere in winter would be comparable to the two to three day errors of modern weather forecasts. Under this project, we are using the Ensemble Filter, and newly gathered surface pressure observations to produce the first-ever reanalysis dataset for the period 1892-2007. This will nearly double the record of 6-hourly tropospheric gridded fields from 60 years to 116, spanning a period for which no gridded upper-air analyses are currently available. These tropospheric circulation fields will also be the first to have objective uncertainty estimates for every variable. Initial results from the reanalysis fields spanning the period 1918-1949 will be presented.