Complex decision-making problem, require large amount of data, and data gathering is one of the most complex tasks in the modeling process. Web services and XML are among the most promising technologies for retrieving data from distributed data sources over the Internet. However, different standards are used for providing spatial and non-spatial data, and data retrieval quality of service attributes vary substantially. The objective of this paper is to propose a spatial data integration architecture, which supports data retrieval from heterogeneous, distributed data sources and accounts for quality of service requirements. Application of the integration architecture is demonstrated using a facility location decision-making problem, which requires such spatial data as distances, customer densities and location of competitors. Computational studies are conducted to evaluate data retrieval time and to identify hidden characteristics of different data sources.