Mendelian randomisation is a technique for assessing causal associations in observational data. Genetic variants associated with the risk factor of interest are regarded in a similar way to random assignment in a clinical trial. The difference in the risk factor due to the genetic variation, however, is materially distinct from the change due to any proposed therapeutic intervention and so might affect the outcome differently. Consequently, it can be misleading to generalise the magnitude of a Mendelian randomisation estimate to the effect of a potential intervention on the risk factor in practice. Awareness of the limitations of such estimates is important for the use of Mendelian randomisation in target based drug development