Background: Excessive antibiotics use increases the risk of resistance. Previous studies have shown that the Centor score combined with Rapid Antigen Detection Test (RADT) for Group A Streptococci can reduce unnecessary antibiotic prescribing in patients with sore throat. According to the former Swedish guidelines RADT was recommended with 2-4 Centor criteria present and antibiotics were recommended if the test was positive. C-reactive protein (CRP) was not recommended for sore throats. Inappropriate use of RADT and CRP has been reported in several studies. Methods: From a larger project 16 general practitioners (GPs) who stated management of sore throats not according to the guidelines were identified. Half-hour long semi-structured interviews were conducted. The topics were the management of sore throats and the use of near-patient tests. Qualitative content analysis was used. Results: The use of the near-patient test interplayed with the clinical assessment and the perception that all infections caused by bacteria should be treated with antibiotics. The GPs expressed a belief that the clinical picture was sufficient for diagnosis in typical cases. RADT was not believed to be relevant since it detects only one bacterium, while CRP was considered as a reliable numerical measure of bacterial infection. Conclusions: Inappropriate use of near-patient test can partly be understood as remnants of outdated knowledge. When new guidelines are introduced the differences between them and the former need to be discussed more explicitly.