The central premise of hydrophobic moment plot methodology is that primary sequence information alone may be used to differentiate between amphiphilic structures which are likely to belong to either surface active or globular proteins. However, the boundaries delineating the regions of the plot appear to be based on information derived from a relatively small protein data set and are not well defined. This, coupled with the uncertainties associated with interpretation of the significance of the hydrophobic moment has led the reliability of this methodology to be questioned with respect to predicting membrane interactions. Accordingly, we have used hydrophobic plot methodology to predict the cellular locations of an expanded set of proteins with known membrane location, including those of the original study. The accuracy of prediction was subjected to chi-square statistical analysis and here we discuss the conclusion that the predictive abilities of this methodology are no better than chance.