Background. Psychiatric diagnoses obtained at personal interview are only moderately reliable and depend critically on accurate self-observation. Reports by family members provide additional information but may be biased. It is unclear how best to combine these two sources of diagnostic data.

Methods. Using complete data on lifetime prevalence for six disorders in ∼1200 male–male twin pairs from a population based registry, we first applied a standard bivariate twin model – which treats self-diagnoses and informant-diagnoses as separate phenotypes – and then examined a ‘multiple-rater’ model – which assumes that self-report and co-twin-report are fallible indices of one underlying disease liability. Best-fit models were chosen using Akaike's information criterion.

Results. Standard bivariate analyses indicated that the same genetic factors accounted for variation in self-reported and co-twin-reported diagnoses. The multiple-rater model produced a substantial decrease in variance attributed to individual-specific environment and a proportional increase in heritability of liability for major depression, generalized anxiety disorder, alcohol dependence and adult antisocial behaviour, but not for drug abuse/dependence or regular tobacco use. The best-fit model consistently included either a ‘bias’ or a ‘correlated error’ path. No evidence for family environmental risk factors was found for any disorder.

Conclusion. The genetic factors that influence self-report psychiatric illness also influence psychiatric illness as described by relatives. For many psychiatric disorders, incorporation of self-report and family history data in a single model may reduce measurement error and increase estimates of heritability. However, account must be taken of the fact that family history reports are systematically biased. While promising, these results are preliminary and require replication.