The uniqueness and limited amounts of forensic samples and samples from objects of cultural heritage together with the complexity of their composition requires the application of a wide range of micro-analytical methods, which are non-destructive to the samples, because these must be preserved for potential late revision. Laboratory powder X-ray micro-diffraction (micro-XRD) is a very effective non-destructive technique for direct phase analysis of samples smaller than 1 mm containing crystal constituents. It compliments optical and electron microscopy with elemental micro-analysis, especially in cases of complicated mixtures containing phases with similar chemical composition. However, modification of X-ray diffraction to the micro-scale together with its application for very heterogeneous real samples leads to deviations from the standard procedure. Knowledge of both the limits and the phenomena which can arise during the analysis is crucial for the meaningful and proper application of the method. We evaluated basic limits of micro-XRD equipped with a mono-capillary with an exit diameter of 0.1 mm, for example the size of irradiated area, appropriate grain size, and detection limits allowing identification of given phases. We tested the reliability and accuracy of quantitative phase analysis based on micro-XRD data in comparison with conventional XRD (reflection and transmission), carrying out experiments with two-phase model mixtures simulating historic colour layers. Furthermore, we demonstrate the wide use of micro-XRD for investigation of various types of micro-samples (contact traces, powder traps, colour layers) and we show how to enhance data quality by proper choice of experiment geometry and conditions.