An improved swarm-based optimization algorithm from the Bees Algorithm family for solving complex optimization problems is proposed. The algorithm performs a form of exploitative local search combined with random exploratory global search. This thesis details the development and optimization of this algorithm and demonstrates its robustness. The development includes a new method of tuning the Bees Algorithm called Meta Bees Algorithm and the functionality of the proposed method is compared to the standard Bees Algorithm and to a range of state-of-the-art optimisation algorithms. A new fitness evaluation method has been developed to enable the Bees Algorithm to solve a stochastic optimisation problem. The new modified Bees Algorithm was tested on the optimisation of parameter values for the Ant Colony Optimisation algorithm when solving Travelling Salesman Problems. Finally, the Bees Algorithm has been adapted and employed to solve complex combinatorial problems. The algorithm has been combined with two neighbourhood operators to solve such problems. The performance of the proposed Bees Algorithm has been tested on a number of travelling salesman problems.