Trace element micro-analysis has evolved steadily since the early days of EPMA, yet remains an extraordinarily challenging subject. The enhanced capabilities of modern instrumentation, including the use of spectrometers with high X-ray collection efficiencies, high brightness electron sources, and improved stability all contribute to our ability to improve detection limits and analytical spatial resolution. Along with much improved software for data acquisition and analysis, recent progress in EPMA has made the trace realm more accessible than ever. High count precision can be obtained in order to easily bring analytical sensitivity into the single ppm range, but accuracy remains the greatest struggle. With the exception of the calibration, all sources of error encountered in major element analysis are magnified in trace analysis, and other sources become apparent where high spatial resolution is needed. Beam damage and charge effects are difficult problems in high sensitivity, high spatial resolution analysis, particularly in the analysis of insulators. Software can minimize some of the resulting effects on count rates during acquisition in order to improve accuracy, and analysts can empirically evaluate the conditions of analysis (count time, voltage, current, etc.) to try to minimize these effects. Trace analysis is fundamentally an exercise in background characterisation, and the acquisition and evaluation of background is a subject of developing methodology. Background curvature and interferences can result in considerable inaccuracy, but can be dealt with via detailed quantitative wavelength scanning or multi-point spectral acquisitions which allow proper regression of the background shape. In the absence of excellent quality trace element secondary standards of similar matrix to unknowns, blank testing and consistency standards can be used to test at least some aspects of the methods employed. Ultimately, the analyst must rely on accuracy evolving from application of the most rigorous protocols.