In this study, trajectory model sensitivity to the input meteorological data and vertical transport method is quantified. ERA-40 and NCEP/NCAR reanalysis data sets are used to compare isentropic and kinematic 3-D tropospheric trajectories. Expanding on previous trajectory sensitivity studies, we accumulate deviation statistics for an entire year and for three geographically diverse sites. The horizontal trajectory deviations are summarized as a percentage of average distance traveled. These results allow ranking from least to greatest among the five causes of trajectory uncertainty investigated here: minor differences in computational methodology, 3--4%; time interpolation, 9--25%; vertical transport method, 18--34%; meteorological input data, 30--40%; and combined two-way differences in vertical transport method and meteorological input data, 39--47%. Although the deviations are somewhat dependent on starting location because of the influence of meteorology, at all three sites, 3-D trajectories attained higher elevations and wind speeds than isentropic trajectories. In addition, deviation statistics for 3-D trajectories exceeded those for isentropic trajectories. The reasons for this derive from uncertainties in the supplied vertical wind fields and the higher wind speeds in 3-D trajectories on the one hand and the vertical constraints imposed by the isentropic assumption on the other.