Abstract Background The aim of this paper is to provide the rationale for an evaluation design for a complex intervention program targeting loneliness among non-institutionalized elderly people in a Dutch community. Complex public health interventions characteristically use the combined approach of intervening on the individual and on the environmental level. It is assumed that the components of a complex intervention interact with and reinforce each other. Furthermore, implementation is highly context-specific and its impact is influenced by external factors. Although the entire community is exposed to the intervention components, each individual is exposed to different components with a different intensity. Methods/Design A logic model of change is used to develop the evaluation design. The model describes what outcomes may logically be expected at different points in time at the individual level. In order to address the complexity of a real-life setting, the evaluation design of the loneliness intervention comprises two types of evaluation studies. The first uses a quasi-experimental pre-test post-test design to evaluate the effectiveness of the overall intervention. A control community comparable to the intervention community was selected, with baseline measurements in 2008 and follow-up measurements scheduled for 2010. This study focuses on changes in the prevalence of loneliness and in the determinants of loneliness within individuals in the general elderly population. Complementarily, the second study is designed to evaluate the individual intervention components and focuses on delivery, reach, acceptance, and short-term outcomes. Different means of project records and surveys among participants are used to collect these data. Discussion Combining these two evaluation strategies has the potential to assess the effectiveness of the overall complex intervention and the contribution of the individual intervention components thereto.