Lessons from past psychiatric genetic research, together with key issues in psychiatry requiring genetic investigation, are reviewed in order to consider the implications for the ways forward. It is argued that traditional quantitative genetics needs to use a combination of twin, adoptee, and family strategies, to examine continuities and discontinuities in psychopathology between childhood and adult life, to compare dimensions and categories, to employ adequate conceptualization and measurement of disorders, to use statistical techniques based on latent constructs, to use biological trait indicators where possible, to examine risk factors as well as diseases, to include good measures of postulated environmental risk variables, to study the interplay between genes and environment, and to study the key assumptions underlying genetic strategies. Molecular cytogenetics needs to consider both the general and specific psychopathological risks associated with chromosome abnormalities and to examine the mechanism involved, to examine the role of submicroscopic chromosomal deletions and of mitochondrial disorders, and to investigate the mechanisms involved in trinucleotide repeat amplifications that take place during intergenerational transmission. Molecular genetics needs to make greater use of smaller pedigrees in view of the concerns over phenotypic definition and genetic heterogeneity in very large extended dense pedigrees, to use sib-pair designs in view of the likelihood that most psychiatric disorder will prove to be multifactorial, to combine association strategies with linkage analyses, to pay careful attention to the definition of phenotypes in probands, to remain in close touch with other branches of biological psychiatry, and to make effective use of collaboration between centers. To date, transgenic models have had a rather limited application in psychiatry but, despite their difficulties, they are likely to provide an underpinning for gene therapy in disorders where that seems feasible.