For better understanding of soot formation and oxidation processes applicable to diesel engines, the size, morphology, and nanostructure of soot particles directly sampled in a diesel spray flame generated in a constant-volume combustion chamber have been investigated using Transmission Electron Microscopy (TEM). For this soot diagnostics, the effects of the sampling processes, TEM observation methodology and image processing methods on the uncertainty in the results have not been extensively discussed, mainly due to the complexity of the analysis. This study particularly aims to reveal the uncertainties due to sampling methods (mainly the effect of shot-by-shot fluctuation of “identical” single-shot spray flames on the sampled soot properties), TEM observation methods (mainly the fluctuation of soot morphology among different on-grid locations and the number of soot particle samples required to compensate for the fluctuations) and analysis methods (mainly operator-by-operator fluctuation of morphology measurements and the effects of averaged diameter and cutoff number for primary particles in aggregates on the determination of fractal dimension). The analysis shows that the shot-by-shot fluctuation of diesel spray combustion does affect the sampled soot density or coverage on the grid, but ultimately the size and morphology distribution of soot particles is consistent, provided there is adequate inspection and analysis of particles on the grid. Although the density, primary and aggregate sizes fluctuate with location on the TEM grid, averaging of adequate number of particlesproduces reliable analysis. The operator dependency in the manual primary size measurements is notable among inexperienced operators, but it is converged by operators' adequate experience.