Background: The mental health burden on primary care is substantial and increasing. Anxiety is a major contributor. Stepped collaborative care (SCC) is implemented worldwide to improve patient outcomes, but long term real-world evaluations of SCC do not exist. Using routinely used electronic medical records from more than a decade, we investigated changes in anxiety prevalences, whether physicians made distinction between non-severe and severe anxiety, and whether these groups were referred and treated differently, both non-pharmacologically and pharmacologically. Methods: Retrospective assessment of anxiety care parameters recorded by 54 general practitioners between 2003 and 2014, in the electronic medical records of a dynamic population of 49,841-69,413 primary care patients. Results: Substantial shifts in anxiety care parameters have occurred. The prevalence of anxiety symptoms doubled to 0.9% and of anxiety disorders almost tripled to 1.1%. Use of ICPC codes seemed comprehensive and use of instruments to support in anxiety level differentiation increased to 13% of anxiety symptom and 7% of anxiety disorder patients in 2014. Minimal interventions were used more frequently, especially for anxiety symptoms (OR 21 [95% CI 5.1-85]). The antidepressant prescription rates decreased significantly for anxiety symptoms (OR 0.5 [95% CI 0.4-0.8]) and anxiety disorders (OR 0.6 [95% CI 0.4-0.8]). More patients were referred to psychologists and psychiatrists. Conclusions: We found shifts in anxiety care parameters that follow the principles of SCC. Future primary care research should comprehensively assess the use of the SCC range of therapeutic options, tailored to patients with all different anxiety severity levels.