Official development agencies are increasingly supporting civil society lobby and advocacy (L&A) to address poverty and human rights. However, there are challenges in evaluating L&A. As programme objectives are often to change policies or practices in a single institution like a Government Ministry, L&A programmes are often not amenable to large-n impact evaluation methods. They often work in strategic partnerships to foster change; hence, contribution may be a more relevant evaluation question than attribution. Small-n qualitative approaches are available to measure the effectiveness of L&A which use the theory of change as their analytical framework. We conducted a meta-evaluation of 36 evaluations of multi-component international programmes to support civil society L&A across Asia, Africa and Latin America, comprising the majority of programmatic support from one international donor. We assessed the confidence in causal claims in the evaluations using a new tool that we developed. Assessments of the contribution of the programmes to the changes in outcomes were not provided in many of the evaluations, nor were predictable sources of bias addressed. Given that L&A programmes are likely to adopt an influencing approach where many different inside-track and outside-track engagement objectives, opportunities and strategies are attempted, many of which might be expected to fail, there appeared to be a clear bias in the evaluations towards reporting outcomes that were achieved, ignoring those that were not. We provide guidance on how to improve the design, conduct and reporting of small-n qualitative evaluations of aid effectiveness.