Abstract Background Type 2 diabetes (T2D) is an increasingly dominant disease. Interventions are more effective when carried out by a prepared and proactive team within an organised system — the integrated care (IC) model. The Chronic Care Model (CCM) provides guidance for its implementation, but scale-up of IC is challenging, and this hampers outcomes for T2D care. In this paper, we used the CCM to investigate the current implementation of IC in primary care in Flanders (Belgium) and its variability in different practice types. Methods Belgium contains three different primary-care practice types: monodisciplinary fee-for-service practices, multidisciplinary fee-for-service practices and multidisciplinary capitation-based practices. Disproportional sampling was used to select a maximum of 10 practices for each type in three Flemish regions, leading to a total of 66 practices. The study employed a mixed methods design whereby the Assessment of Chronic Illness Care (ACIC) was complemented with interviews with general practitioners, nurses and dieticians linked to the 66 practices. Results The ACIC scores of the fee-for-service practices — containing 97% of Belgian patients — only corresponded to basic support for chronic illness care for T2D. Multidisciplinary and capitation-based practices scored considerably higher than traditional monodisciplinary fee-for-service practices. The region had no significant impact on the ACIC scores. Having a nurse, being a capitation practice and having a secretary had a significant effect in the regression analysis, which explained 75% of the variance in ACIC scores. Better-performing practices were successful due to clear role-defining, task delegation to the nurse, coordination, structured use of the electronic medical record, planning of consultations and integration of self-management support, and behaviour-change intervention (internally or using community initiatives). The longer nurses work in primary care practices, the higher the chance that they perform more advanced ...