The chi-square test has been extensively employed by geologists and geomorphologists as a measure of the strength or statistical significance of preferred trends within orientation data. The x 2 value obtained is dependent on arbitrary selection of the pattern of sectors into which orientation measurements are grouped. The calculation of all possible x 2 solutions for 98 sets of 50 stone orientation measurements demonstrated that a wide range of x 2 values could be calculated for most samples. The hitherto widespread use of the test in this way cannot therefore be considered a valid mode of analysis, and the results of previous studies employing this methodology must be treated with extreme caution. The use of a different test statistic is advocated.