Mendelian Randomization (MR) is an analytical method of statistical genetics that uses genetic variants as instrumental variables (IVs) to infer whether an observational association between a particular exposure and an outcome of interest is consistent with a causal effect. A genetic variant is a valid IV if (a) it is associated with the exposure (relevance condition), (b) there are no associations between the IV and any confounders (independence condition), and (c) the IV has no direct effect on the outcome (exclusion-restriction condition). Violation of the IV assumptions may lead to invalid conclusions from MR results. Robust methods for MR analysis have been developed to provide reliable causal estimates when IV assumptions are violated, and sensitivity analyses using robust methods are recommended to detect assumption violations. Simulation studies have been performed to investigate which types of violation of IV assumptions are adequately addressed by different MR methods.