If paleomagnetic poles from individual studies are compared with mean paleomagnetic poles, it is found that there is significant numbers of deviations which are greater than would be expected when the formal statistical error associated with the position of the individual pole is considered. This suggests that in many cases, individual poled have not completely samples the earth's magnetic field and its variation to give a good estimate of the mean field. This further suggests that uniform weighting the individual studies to obtain a mean pole is incorrect. We develop a method of determining mean poles in which the information content of individual studies is used as a weighting factor. No assumptin need be made that the individual studies represent a mean pole for the time period under consideration. Studies in which it is known that the mean pole has not been adequately determined may then be used in the analysis by giving such studies a low weighting factor in the analysis. We also introduce a method of weighting dependent on the age overlap between the individual study and the age window for which a mean pole is to be calculated. The most important determinant of the polar wandering path is, however, the data set used.