Despite the growing literature that collaboration is a 'good' thing, there are calls emphasising the need for evidence of its effectiveness. However, the nature of the evidence to assess effectiveness is less clear. This paper examines the components that contribute to the challenges that confront evidence on collaboration. It considers the differing interpretations that have been placed on evaluation and explores how ways of determining the outcomes of collaboration and the levels of outcome measurement to assess collaborative effectiveness are influenced by the multifactorial nature of the concept. Evidence on the impact of collaboration is influenced by the diversity of perspectives and conceptual facets, and difficulty in measurement of the notions involved. Other factors discussed are the choice of macro or micro evaluation, of proximal or distal indicators, of short and long-term effects, or of individual-level or collective community-level outcomes. The suitability of randomised controlled trials for the measurement of collaborative outcomes as well as the requirement of mixed methods evaluations are highlighted. An evaluation of five community partnerships in South Africa is employed as an example to link the evaluation concepts that are discussed to a real enquiry. If collaboration is to be successful in making a difference in the lives of people, then increasing the precision and context of appraising its effectiveness will reduce the nature of inconclusive evidence and is likely to improve the practice of partnerships, coalitions and joint working in health and social care.