Correlation is of direct importance to all engineers-both those who generate new information and those who wish to make use of it. The solution of theoretical models to predict behavior does not preclude the need for correlation since the output is generally a set of discrete values that differ from experimental data only by virtue of their precision and regularity.; The methodologies that lead to improved correlating equations are examined in general and by means of specific examples in fluid now and heat transfer. The identification of the optimal dimensionless groupings and the derivation and incorporation of theoretically based asymptotes are shown to be the keys to better correlation. Nonconformity to known theoretical relationships and constraints and the uncritical examination of experimental data or computed values are found to be the principal sources of miscorrelation. Correlations should be expected to evolve and to be replaced as new data are obtained and new theoretical relationships are derived. However, out of inertia and nostalgia, obsolete correlations are found to persist in our textbooks, in our handbooks, and even in the software for computer-aided design.