Obesity (or adiposity more generally) has well characterised relationships with various intermediate measures of cardiovascular health and endpoints when assessed observationally. Whilst these have been derived from large and well undertaken studies, they of course suffer a series of known limitations which lead the community to need alternative approaches if interested in testing causality more directly. Trials of interventions reducing levels of adiposity, studies in differing confounding frameworks and prospective examination are all sensible approaches to dealing with this difficult problem, however to this has been recently added Mendelian randomisation. As part of a battery of methods used to help situations where causality is difficult to assess, the use of genetic variation reliably associated with a risk factor of interest (in this case obesity/adiposity/BMI) to act as a proxy measure for it can allow for the reassessment of causality and the generation of useful, clinically applicable, information. We have undertaken a series of studies across numerous data collections which have examined the causal impact of adiposity (measured via BMI) on cardiovascular outcomes and intermediates of cardiovascular health. Here I will summarise pertinent results and consider the implications of these for the consideration of adiposity as a risk factor and of BMI as a target for improving health.