Smoking has consistently been associated with poor mental health but determining the causal nature and direction of this association is challenging. Observational studies may be biased by confounding and/or reverse causation, and randomized controlled trials (RCTs) are not feasible for the relationships under study. As an alternative causal inference method, Mendelian randomization (MR) is becoming increasingly popular. MR utilizes large genetic datasets and is, to some extent, analogous to a RCT. Instead of participants being assigned to experimental conditions, MR compares subgroups in the population which are at differing levels of genetic risk for a proposed risk factor. Because of random assortment at meiosis, genetic variants should not be associated with confounders, allowing less biased causal inference. First, we will present a systematic review of MR studies which found robust evidence for a bidirectional, increasing relationship between smoking and (symptoms of) mental disorders. Second, we will present more recent MR research, which investigated the relationship between smoking and brain structure and found evidence that smoking causally decreases subcortical brain volume. We will discuss the possibility that the effects of smoking on brain structure could be a mediating pathway between smoking and mental disorders and finish with recommendations for future MR studies.