Diabetes education or self-management programmes are complex interventions. Their evaluation is difficult because of problems in identifying and separately assessing the effect of the various components of the intervention. A phased approach defining sequential stages of a continuum of increasing evidence has been proposed as a framework for the design and evaluation of such complex interventions. As an example we present the available evidence for diabetes treatment and teaching programmes implemented in Germany. Evidence is compiled for structured group treatment and teaching programmes for Type I diabetes, non-insulin dependent Type II diabetes, and hypertension according to the following sequential stages of increasing evidence: (i) preclinical or theoretical phase; (ii) modelling the components of the intervention; (iii) exploratory trials; (iv) randomized controlled trials; (v) phase of implementation including replication and transfer to different settings. Evidence for most of these phases has been generated for the three programmes, although individual studies do not fulfill all important quality criteria by today's standards. The time span for gathering the evidence from the theoretical phase to surveillance after implementation was about 20 years. It can only be speculated which parts of the programmes are the most active ones. The presentation of a continuum of increasing evidence for diabetes education or self-management programmes could provide useful information for the appraisal of such complex interventions. Since this evidence cannot be readily extracted from databases we suggest that other research groups present their data in a similar way.