Introduction. Aseptic loosening, the clinical endpoint of osteolysis, remains the leading cause of total hip arthroplasty (THA) failure, and is caused by a host response to wear debris that varies between individuals. Although several candidate gene studies have identified loci associated with osteolysis susceptibility, there have been no systematic studies at genome-wide level. We aimed to identify risk loci associated with osteolysis by conducting a genome-wide association study. Methods. 3,706 Caucasian European patients following THA were studied. The discovery cohort comprising 894 patients (317 with osteolysis) were genotyped using the Illumina-610 beadchip followed by 1000 Genome-based imputation covering 10 million single nucleotide polymorphisms (SNPs). Phenotypes were transformed to normality where required, regressed on important covariates and z-standardised. Following quality control, osteolysis case-control analysis and a quantitative trait association analysis for time to prosthesis failure were undertaken. Index SNPs p<9×10. −4. were taken forward for replication in a second cohort comprising 2,812 subjects (834 osteolysis cases) recruited from the Norwegian arthroplasty registry. Genotyping was undertaken using Sequenom MassARRAY iPLEX Gold assay and association analyses undertaken using logistic and linear regression. Summary statistics were combined in a fixed-effects meta-analysis framework. Results. The strongest signal associated with time to prosthesis failure lay within DEFB129 gene. The signal index SNP, rs6105394, approached genome wide significance at p=5.75×10. −7. Two signals in the susceptibility analysis also approached genome-wide significance, 1 within CAMK4 (rs306105, OR 0.41, p=6.54×10. −7. ) and 1 upstream of PLNXA2 (rs11119057, OR 0.96, p=6.44×10. −7. ). Following meta-analysis, the strongest signal in the susceptibility analysis remained that within CAMK4 (rs306105, p=3.79×10. −4. ). The strongest signal associated with time to failure was just upstream of CNTN3 (rs1374879, p=2.15×10. −5. ). Discussion. We have identified promising loci associated with osteolysis and time to prosthesis failure although not at genome-wide significance (p<5×10. −8. ). In order to further validate these loci, larger genome wide association analysis is required