Risks and potential benefits of distributing software development projects globally depend to a large extent on how to allocate work to different development sites and regions. Existing methods in task allocation are likely to omit the relevance of considering a multitude of criteria and the impact of task distribution on risks and potential benefits. To assess risks stemming from specific work distributions and to exploit organization-specific experience, we have developed a customizable risk-driven model. It consists of two main steps: Suggesting a set of task allocation alternatives based on project- and site-specific characteristics and analyzing it with respect to possible project risks stemming from the work distribution. To evaluate the model, we conducted a series of semi-structured interviews in a multinational IT company. The results of the evaluation show that the suggestions of the model mostly comply with the retrospective views voiced by the involved experienced managers.