Rationale: Idiopathic pulmonary fibrosis (IPF) is a complex lung disease characterised by scarring of the lung that is believed to result from an atypical response to injury of the epithelium. Genome-wide association studies have reported signals of association implicating multiple pathways including host defence, telomere maintenance, signalling and cell-cell adhesion. Objectives: To improve our understanding of factors that increase IPF susceptibility by identifying previously unreported genetic associations. Methods and measurements: We conducted genome-wide analyses across three independent studies and meta-analysed these results to generate the largest genome-wide association study of IPF to date (2,668 IPF cases and 8,591 controls). We performed replication in two independent studies (1,456 IPF cases and 11,874 controls) and functional analyses (including statistical fine-mapping, investigations into gene expression and testing for enrichment of IPF susceptibility signals in regulatory regions) to determine putatively causal genes. Polygenic risk scores were used to assess the collective effect of variants not reported as associated with IPF. Main results: We identified and replicated three new genome-wide significant (P<5×10−8) signals of association with IPF susceptibility (associated with altered gene expression of KIF15, MAD1L1 and DEPTOR) and confirmed associations at 11 previously reported loci. Polygenic risk score analyses showed that the combined effect of many thousands of as-yet unreported IPF susceptibility variants contribute to IPF susceptibility. Conclusions: The observation that decreased DEPTOR expression associates with increased susceptibility to IPF, supports recent studies demonstrating the importance of mTOR signalling in lung fibrosis. New signals of association implicating KIF15 and MAD1L1 suggest a possible role of mitotic spindle-assembly genes in IPF susceptibility.