We use pattern matching of experimental and dynamically simulated backscattered Kikuchi diffraction (BKD) patterns to increase the orientation precision of electron backscatter diffraction (EBSD) measurements. In order to quantify the improvement in orientation precision, we analyze the experimental distribution of the kernel average misorientation (KAM) angles. We find that for the same raw data, i.e. the same EBSD data acquisition time budget, the pattern matching approach improves the KAM resolution by an order of magnitude compared to orientation data delivered from the conventional Hough-transform based data analysis. This quantitative improvement enables us to interpret small orientation changes in plastically deformed materials which are hidden in noisy orientation data delivered from the reference EBSD system. As an application example, we analyze a ferritic-martensitic steel (P92) sample before and after low-cycle fatigue (LCF) loading (at 620 °C. Whereas the low precision of the EBSD orientation data from the manufacturer software does not allow a reliable discrimination of the gradually changing microstructure, we find very clear systematic differences of the local microstructure after the pattern matching orientation refinement of the initial, raw pattern data. For the investigated P92 sample, the KAM-angle histograms are well described by two log-normal distributions indicating the already tempered and the remaining and mostly untempered martensite.