De novo donor-specific human leucocyte antigen (HLA) antibodies (dnDSA) are associated with increased risk of rejection and mortality in solid organ transplantation. Such dnDSA is produced in some recipients upon allorecognition of mismatched HLA posttransplant. HLA matching is not currently considered in the allocation of deceased donor hearts and lungs and pre-transplant immunological risk stratification is based entirely on the mean fluorescence intensity (MFI) of circulating donor-directed HLA antibodies. HLA epitope-based matching tools predict B-cell or T-cell HLA epitopes that are present in the donor's HLA but absent in the recipient's HLA. We hypothesized that patients with higher epitope mismatch loads would be at increased risk of dnDSA development. We retrospectively analysed 73 heart and/or lung transplant recipients who were tested for DSA between 2015 and 2020. HLAMatchmaker, PIRCHE-II and HLA epitope mismatch algorithm (HLA-EMMA) were used to calculate eplet mismatch (EpMM) loads, T-cell epitope mismatch (TEpMM) loads and solvent accessible amino acid mismatch (SAMM) loads, respectively. Multivariate analyses showed that HLA-EMMA was the only tool with a significant association between the total score for all HLA loci and dnDSA production [odds ratio (OR) 1.021, 95% confidence interval (CI) 1.003-1.042, p = .0225] though this increased risk was marginal. The majority of dnDSA were directed against HLA-DQ and patients with higher HLA-DQ TEpMM loads (OR = 1.008, CI = 1.002-1.014, p = .007), and HLA-DR+DQ SAMM loads (OR = 1.035, CI = 1.010-1.064, p = .0077) were most at risk of producing dnDSA.We also showed that patients with a risk epitope within the HLA molecule encoded for by HLA-DQA1*05 + HLA-DQB1*02/03:01 were significantly more likely to produce dnDSA. The use of HLA epitope-based matching tools could be used for cardiothoracic transplant risk stratification to enable early intervention and monitoring of patients at increased risk of producing dnDSA.