Sleep disturbances, encompassing both sleep disorders and self-reported sleep difficulties/tiredness, have been associated with blood lipid traits (Noordam et al., 2019) and cardiometabolic disease risk (Cappuccio & Miller, 2017). It is unclear to what extent these associations are causal in nature, and if they are, the direction of causality. To help clarify this, we conducted Mendelian randomisation (MR) analyses on a range of sleep traits and lipid biomarkers using summary statistics from the UK Biobank and 23andMe cohorts. The number of lipid and sleep traits studied was reduced by choosing representative traits amongst highly correlated clusters of traits. Additionally, traits with non-significant estimates of heritability were excluded and we focused on a sub-set of remaining trait pairs with nominally significant genetic correlations (n=108). Lipid and sleep traits are known to be mediated by the circadian system, and so in the first instance we performed bidirectional analyses using two methods (LHC-MR, MR-CAUSE) that account for horizontal correlated pleiotropy. TwoSampleMR methods were then used as sensitivity analyses. We identified a total of nine trait pairs with FDR-significant evidence for a causal relationship using both LHC-MR and MR-CAUSE, with eight in the sleep-to-lipid direction and one in the lipid-to-sleep direction. In general, TwoSampleMR results were not in agreement with LHC-MR and MR-CAUSE, potentially because the former methods do not model horizontal correlated pleiotropy. Overall, our results provide modest evidence for a causal effect of selected sleep traits on lipid metabolism. However, lack of power in the present study means that we cannot rule out weak causal effects for other trait pairs. Future MR analyses leveraging larger GWAS datasets, such as the Global Lipids Genetic Consortium, are planned to validate and extend these findings.