Plot location errors (distance between recorded and true coordinates) of National Forest Inventory (NFI) cause a mismatch between plot data and remotely sensed data and might prevent re-measurement because plots with large location errors are hard to find. However, few studies have examined the detailed distribution of errors and related factors, and no studies have verified whether the large errors prevent re-measurement. This study analyzed data related to 101 plots in central Japan that had been established by the First NFI (1999–2003) on a 4-km grid. Plot location errors of these plots were measured by revisiting. Selective Availability (SA) of the Global Positioning System, which degraded location accuracy until May 2000, was the most important factor in increasing plot location errors. The mean errors were 58.6 and 15.0 m, with and without SA, respectively. In 12 plots with large plot location errors (the mean error was 84.6 m), re-measurement in the Second NFI (2004–2008) was not conducted because plot locations could not be found. In these situations, alternative new plots were established; however, their species compositions were significantly different from the initial NFI plots. Plot location errors of NFI adversely affect the continuity of plot data as well as the analysis with remotely sensed data.