Despite significant advances in pre- and postnatal care over the last century, adverse pregnancy related events still occur frequently. Birthweight, a proxy for intrauterine growth, acts as an important predictor of newborn and infant survival and has been observationally associated with future offspring cardiometabolic risk. Recent years have seen an increase in publicly available large-scale proteomic genome wide association studies (GWAS). Typically, cis-acting genetic variants associated with protein levels display larger effect sizes than those associated with common complex traits. As such, large-scale protein GWAS could provide a useful adjunct and powerful discovery tool for improved understanding of the molecular pathways underlying common complex traits. We used Mendelian randomisation (MR) to investigate potential causal effects of the mother’s and fetal blood proteome on pregnancy related outcomes including birthweight, placental weight, sporadic miscarriage, and stillbirth. We generated a list of genetic instruments to act as proxies for plasma proteins by combining two recent large protein GWAS (4,719 proteins N=35,559 individuals; 4,775 proteins N=10,708 individuals). We identified 1,711 proteins with valid cis-pQTLs for use as genetic instruments. We identified robust causal associations (MR p-value < 2.92x10-5 and evidence of colocalisation, posterior probability > 0.8) involving fetal effects and/or maternal effects on birthweight. Increased levels of PSG7 and TNFRSF17 and decreased levels of INHBB, PLCG1, and UBASH3B in the fetal circulation were causally associated with increased birthweight. Similarly, increased levels of LIMA1 and decreased levels of LGALS4 in the maternal circulation were causally associated with increased birthweight. We did not find any significant effects of proxied maternal or fetal proteins on gestational age, placental weight, sporadic miscarriage, or stillbirth, perhaps due to the smaller size of their GWAS meta-analyses. Our results implicate several proteins that may be involved in the aetiology of birthweight and that will need to be replicated in independent datasets.