In a component-based development process the selection of components is an activity that takes place over multiple lifecycle phases that span from requirement specifications through design to implementation and integration. In different phases, different assumptions are valid and different granularity of information is available, which has a consequence that different procedure should be used in the selection process and an automated tool support for an optimized component selection would be very helpful in each phase. In this paper we analyze the assumptions and propose the selection procedure in the requirements phase. The selection criterion is based on cost minimization of the whole system while assuring a certain degree of satisfaction of the system requirements that can be considered before designing the whole architecture. For the selection and optimization procedure we have adopted the DEER (DEcision support for componEnt-based softwaRe) framework, previously developed to be used in the selection process in the design phase. The output of DEER indicates the optimal combination of single COTS (Commercial-Off-The-Shelf) components and assemblies of COTS that satisfy the requirements while minimizing costs. In a case study we illustrate the selection and optimization procedure and an analysis of the model sensitivity to changes in the requirements.