The main strength of the Rietveld method as a tool for quantitative phase analysis lies in its ability to control and correct various parameters which influence powder diffraction patterns. The main sources of errors in this type of analysis are due to differences in properties of various constituents of a mineral mixture and the inability to model these consistently. Sources and products of acid mine drainage represent a system where these differences are especially prominent and therefore pose a problem for quantitative phase analysis. Large differences in hardness, mineral habit, and absorption of X-rays plus frequent amounts of poorly crystalline and amorphous components are the main reasons for erroneous results in this kind of samples. We have used a combination of procedures to overcome these difficulties: stepwise grinding of samples to avoid amorphisation of very soft components in the presence of very hard ones, preparation of moderately pressed samples to diminish porosity and surface roughness, application of numerical corrections for the absorption contrast and preferred orientation and the application of a specific addition method for the determination of unknown components in the mixture. The procedures should be tested on synthetic mixtures for each specific instrument/laboratory and type of analysis, as a standard part of the quantitative phase analysis and the necessary way for the estimation of expected errors. In our case mixtures of pyrite, dolomite, gypsum and amorphous silica were applied as proper models. The procedures developed on test mixtures result in estimated errors of determination typically lower than 2 wt%. The approach was applied to the analysis of the complex problem of estimating mass balance in oxidized mine tailings from the Nanisivik Mine, Nunavut, Canada revealing an oxidized proportion of 9 % of total pyrite content in the well drained tailings during a period of 8 years and a negligible oxidation in the water-covered tailings.