The traditional role of twin studies has been to assess the relative role of genetic factors as a first step in defining the genetic architecture of complex traits. This has been based on the realization that monozygotic pairs (MZ) share all their genes, while dizygotic pairs (DZ) share 50% of their genes on average. Thus, greater similarity of MZ pairs compared to DZ pairs has been taken as prima facie evidence of the role of genetic factors. This is true provided the environmental similarity of MZ pairs is not greater than for DZ pairs for effects relevant to the trait in question. This first step in genetic studies was carried out long ago in many research areas, but not in others. More detailed knowledge of the genetic architecture of traits is then obtained by other means. In this paper, we give a brief overview of some results for metabolic diseases (ischaemic heart disease, hypertension, subarachnoid haemorrhage, NIDDM and IDDM) using the classical twin approach in a large, unselected population-based twin cohort. We also outline approaches to using twins that we believe will continue to be useful, particularly for the study of environmental effects.