Hemagglutinin (HA), a variable viral surface protein, is essential for influenza vaccine development. Annually, traditional trivalent vaccines containing influenza A/H1N1, A/H3N2 and B viruses are administered globally, which are not very effective for the mutations in HA protein. The aim of this study was to design a multi-epitope vaccine containing epitopes of the HA protein of H1N1, H3N2 and B viruses using immunoinformatics methods. The HA protein epitope prediction was performed using Immune Epitope Database. Toxicity, antigenicity and conservancy of the epitopes were evaluated using ToxinPred, VaxiJen and Epitope Conservancy Analysis tools, respectively. Then, nontoxic, antigenic and high conserved epitopes with high prediction scores were selected. Their binding affinity was evaluated against human and mouse MHC class I and II molecules using the HPEPDOCK tool. Physicochemical properties and post-translational modifications were evaluated using ProtParam, SOLpro and MusiteDeep tools, respectively. Top selected epitopes were joined using linkers to produce the best effective recombinant trivalent vaccine candidate to elicit cellular and humoral immune responses in mouse and human host models. These sequences were modeled and verified. By evaluating the results of various analyses of all models and the most similarity to the native HA protein, model 5 was selected as the best model. Finally, in silico cloning of this model as vaccine candidate was performed in pET21. This study was a computer-aided analysis for a multi-epitope trivalent recombinant vaccine candidate against influenza viruses. The efficiency of our best model of vaccine candidates should be validated using in vitro and in vivo studies. Communicated by Ramaswamy H. Sarma