Background Guideline adherence is generally high in Dutch general practices. However, the prescription of insulins to type 2 diabetes mellitus patients is often not in line with the guideline, which recommends NPH insulin as first choice and discourages newer insulins. This qualitative study aimed to identify the reasons why primary care healthcare professionals prescribe insulins that are not recommended in guidelines. Methods Digital focus groups with primary care practitioners were organised. A topic list was developed, based on reasons for preferred insulins obtained from literature and a priori expert discussions. The discussions were video and audio-recorded, transcribed verbatim and coded with a combination of inductive and deductive codes. Codes were categorized into an existing knowledge, attitudes and behaviour model for guideline non-adherence. Results Four focus groups with eleven general practitioners, twelve practice nurses, six pharmacists, four diabetes nurses and two nurse practitioners were organised. The prescription of non-recommended insulins was largely driven by argumentation in the domain of attitudes. Lack of agreement with the guideline was the most prominent category. Most of those perspectives did not reflect disagreement with the guideline recommendations in general, but were about advantages of non-recommended insulins, which led, according to the healthcare professionals, to better applicability of those insulins to specific patients. The belief that guideline-recommended insulins were less effective, positive experience with other insulins and marketing from pharmaceutical companies were also identified as attitude-related barriers to prescribe guideline-recommended insulins. One additional category in the domain of attitudes was identified, namely the lack of uniformity in policy between healthcare professionals in the same practice. Only a small number of external barriers were identified, focusing on patient characteristics that prevented the use of recommended insulins, the availability of contradictory guidelines and other, mostly secondary care, healthcare providers initiating non-recommended insulins. No knowledge-related barriers were identified. Conclusions The prescription of non-recommended insulins in primary care is mostly driven by lack of agreement with the guideline recommendations and different interpretation of evidence. These insights can be used for the development of interventions to stimulate primary care practitioners to prescribe guideline-recommended insulins.