Optimum experimental conditions were determined for X-ray diffraction analysis of the calcite/dolomite ratio of limestone. An equation was derived for determining the number of counting periods necessary to achieve a given level of accuracy. X-ray diffraction peak intensities were measured over the 2θ scan, scanning at a rate of 2 deg/min with counting periods of 40 sec. For greater accuracy, peak areas were measured rather than peak height. Ultra-fine grinding of specimens in a porcelain impact ball mill gave more reproducible results than other specimen-preparation methods, even though the fine grinding caused line-broadening. Mounting the specimen in a holder and rotating it during testing further reduced variance about 1/3. X-rays from a medium-wavelength tube such as copper, a beam-slit width of 3 deg, and a detector-slit width of 0.1 deg were found most suitable. An equation was developed to relate the coefficient of variation to peak and background counts. The calculated coefficient of variation is the maximum variation that may be attributed to the counting statistic, excluding experimental error. By use of the equation or graphs, the minimum coefficient of variation is predicted from one fast scan, and the number and optimum arrangement of additional counting periods to reduce variation to the desired limit can be obtained.