•An undisturbed probe model describing depth-sectioning STEM valid for all STEM techniques is derived under three well-defined conditions.•The model provides closed form expressions for depth-sectioning STEM and enables analysis of the 3D resolution.•The range of validity of the undisturbed probe model is tested using multi-slice simulations.•A comparison of depth sectioning using iDPC- and ADF-STEM is made both in simulations and in experiments and the differences are explained.•Practical criteria for deciding whether a sample is thick or thin are proposed. We derive a model that describes 3D volume imaging in depth-sectioning STEM that is valid for all STEM techniques under three well-defined conditions: linearity, undisturbed probe and elastic scattering. The resulting undisturbed probe model generalizes the widely used idea that the undisturbed probe intensity in three dimensions can be used as the point spread function for depth-sectioning ADF-STEM to all STEM techniques including (A)BF- and iDPC-STEM. The model provides closed expressions for depth-sectioning STEM, which follow directly from the 2D expressions for thin samples, and thereby enables analysis of the 3D resolution. Using the model we explore the consequences of the resulting 3D contrast transfer function (CTF) for the z-resolution at different length scales and illustrate this with experiments. We investigate the validity and limitations of the model using multi-slice simulations showing that it is valid and quantitatively accurate for relatively thick amorphous samples but not for crystalline samples in zone-axis due to channeling. We compare depth-sectioning in iDPC- and ADF-STEM and show that iDPC-STEM can extract information from deeper into the sample, all the way till the bottom of the sample, thereby effectively allowing a thickness measurement. Also the difference in optimal focus conditions between iDPC- and ADF-STEM is explained. Finally, we propose practical criteria for deciding whether a sample is thin or thick.