Recent publications have shown that general aviation accident rates in the United States are decreasing, though they remain much higher than other segments of aviation. What is behind this safety improvement? Are all types of accidents decreasing or are certain types of accidents driving the overall decrease? This study provides the preliminary results of an in depth analysis of the causes of fatal general aviation accidents from 1992 through 2002. First a database of all fatal Part 91 accidents during this timeframe was created by examining the NTSB accident report for each accident and determining the initial cause of the accident. Forty-four different cause categories were used and 1305 individual accidents were analyzed. Both the absolute and relative change of the causes of Part 91 fatal accidents from 1992 through 2002 were considered. This study provides the findings from four years from this time period (1992, 1997, 1999, and 2001) in order to evaluate initial results and trends and to develop some initial hypotheses for later testing. Preliminary analyses of the data reveal that the distribution of fatal general aviation accidents by flight phase differs from commercial aviation accidents. The data also reveals that while pilot error remains a significant cause of fatal accidents, the rate of pilot error accidents is decreasing and CFIT accidents, which are thought to be a particular area of concern within general aviation, are actually not a significant cause of accidents within this sector.