In this paper well defined test samples are used to show how the analysis depth and the matrix material influence the depth resolution of nuclear reaction analysis. The reaction D(3 He, p)α is used to detect. The influence of the atomic number and the overlayer thickness on the depth resolution is studied by covering 10 nm thin deuterated amorphous carbon (a-C:D) films on silicon with tungsten (Z W =74) and titanium (Z Ti =22) of various thicknesses between 500 nm and 8 μm. The most probable depth profiles are calculated from the experimental data with the program NRADC, which implements Bayesian statistics. The resulting apparent layer width of the deuterium containing layer broadens with increasing thickness of the coating and this broadening is more pronounced for coatings with higher Z. These apparent layer widths are a measure for the achievable depth resolution. They are in good agreement with depth resolutions calculated with RESOLNRA. To investigate the Preprint submitted to Elsevier 8 November 2012 depth resolution of the p(15 N, α, γ) 12 C reaction, a 12 nm thin hydrogenated amorphous carbon (a-C:H) film on silicon and a pure tungsten sample are analysed. The width of the instrument function of this method is deduced from the surface hydrogen peak of the pure tungsten sample. The depth resolutions of the two NRAmethods are compared.