The purpose of this study was to examine factors which may affect the reliability of interpreting scree plots. Six experienced and 11 novice raters were each given 10 scree plots; 5 were taken from Cattell and Vogelmann's (1977) paper and reflected artificial data, and 5 were from real data sets. Plots were drawn to simulate computer output. Novices were slightly more reliable than experts; and both were more reliable with artificial data (intraclass correlations of .33 and .33, respectively) than with real data (intraclass correlations of .05 and .00). Eliminating the single Cattell and Vogelmann plot which had error built in resulted in higher reliabilities (.42 and .36). Thus, the reliability of interpreting plots appears poor under conditions which approximate how they are usually read: data which contain error and are plotted on graphs which have a varying ratio of X- to Y-axis scales.