BackgroundMedically unexplained symptoms (MUS) and somatoform disorders are common in general practices, but there is evidence that general practitioners (GPs) rarely use these codes. Assuming that correct classification and coding of symptoms and diseases are important for adequate management and treatment, insights into these processes could reveal problematic areas and possible solutions. Our study aims at exploring general practitioners’ views on coding and reasons for not coding MUS/somatoform disorders.MethodsWe invited GPs to participate in six focus groups (N = 42). Patient vignettes and a semi-structured guideline were used by two moderators to facilitate the discussions. Recordings were transcribed verbatim. Two researchers analyzed the data using structuring content analysis with deductive and inductive category building.ResultsThree main categories turned out to be most relevant. For category a) “benefits of coding” GPs described that coding is seen as being done for reimbursement purposes and is not necessarily linked to the content of their reference files for a specific patient. Others reported to code specific diagnoses only if longer consultations to explore psychosomatic symptoms or psychotherapy are intended to be billed. Reasons for b) “restrained coding” were attempting to protect the patient from stigma through certain diagnoses and the preference for tentative diagnoses and functional coding. Some GPs admitted to c) “code inaccurately” attributing this to insufficient knowledge of ICD-10-criteria, time constraints or using “rules of thumb” for coding.ConclusionsThere seem to be challenges in the process of coding of MUS and somatoform disorders, but GPs appear not to contest the patients’ suffering and accept uncertainty (about diagnoses) as an elementary part of their work. From GPs’ points of view ICD-10-coding does not appear to be a necessary requirement for treating patients and coding might be avoided to protect the patients from stigma and other negative consequences. Our findings supply a possible explanation for the commonly seen difference between routine and epidemiological data. The recent developments in the DSM-5 and the upcoming ICD-11 will supposedly change acceptance and handling of these diagnoses for GPs and patients. Either way, consequences for GPs’ diagnosing and coding behavior are not yet foreseeable.Electronic supplementary materialThe online version of this article (10.1186/s12875-018-0812-8) contains supplementary material, which is available to authorized users.