Abstract : In this thesis, optimum design algorithm is developed for integral abutment bridges. These types of bridges consist of girders, deck slabs that is monolithically casted with abutment and abutment is supported on row of piles. They are joint-less and provide reduction in the maintenance cost. The optimum design algorithm presented considers total cost of the bridge as objective function to be minimized and takes deck thickness, diameter of top and bottom reinforcement in the deck and their spacing, girder sections from AASHTO and its spacing, abutment thickness, vertical and horizontal steel bar diameters, and their spacing, H-steel piles designation and the spacing between these piles as design variables. The design variables need to have discrete integer values due to practical reasons. The design constraints are implemented from AASHTO-LRFD code. There are 26 design constraints altogether and some of which cover flexural resistance of the deck for positive and negative moments, crack control, distribution reinforcement, girder design, abutment design, pile design and serviceability requirements. The mathematical formulation of the design problem turns out to be discrete nonlinear integer programming problem. Mathematical programming techniques are not capable of obtaining the optimum solution of such discrete optimization problems. The only option available is to use soft computing techniques, which are also known as metaheuristics. In this study, five different metaheuristics are employed to attain the solution of the above- described programming problem. These are artificial bee colony algorithm (ABC), the differential evolution optimization algorithm (DE), the teaching and learning-based optimization alg orithm (TLBO), the enhanced beetle antenna search (eBAS) and the grey wolf optimizer algorithm (GWO). The reason several metaheuristics are utilized to determine the solution is because their performance is problem dependent. One of the objectives of the study is to determine which one performs the best among all. All these algorithms require the response of the bridge under several loading conditions, which is utilized to compute the new values of design variables for the next iteration. This is achieved by using application-programming interface (API) of SAP2000. All the metaheuristics are coded in MATLAB and the response of the bridge is transferred to MATLAB program through API. It is noticed that all the five metaheuristics attains the optimum solutions except ABC. The optimum solution obtained by eBAS, DE and TLBO was the same, which is equal to 370,073 (BD) while the solution obtained by GWO was slightly higher by around 1%. TLBO has the fastest convergence and others converged after further iterations but all were converged almost in the same range. In addition, eBAS was the fastest algorithm in terms of running time.