The authors describe the application of multivariate analysis to the problem of estimating intra-family correlations and testing them for statistical significance. This is illustrated by a re-analysis of survey data collected by Miall and Oldham (Clin Sci 1958;17:459-87) on the familial aggregation of blood pressure. The multivariate analysis provides collective tests of significance for parent-child and child-child correlations based on likelihood ratio theory, and maximum likelihood estimates for individual intra-family correlations.