Despite the substantial disease burden of anxiety disorders, only limited or conflicting data on prognostic factors is available. Most studies include patients in the secondary healthcare sector thus, the generalizability of findings is limited. The present study examines predictors of symptom reduction and remission in patients with anxiety disorders in a primary care setting. 214 patients with anxiety disorders, recruited as part of the Collabri Flex trial, were included in secondary analyses. Data on potential predictors of anxiety symptoms at 6-month follow-up was collected at baseline, including patient characteristics related to demography, illness, comorbidity, functional level, life quality, and self-efficacy. The outcomes were symptom reduction and remission. Univariate and multivariate linear and logistic regression analyses were conducted to assess the associations between predictor variables and the outcome, and machine-learning methods were also applied. In multiple linear regression analysis, anxiety severity at baseline (β = -6.05, 95% CI = -7.54,-4.56, p < 0.001) and general psychological problems and symptoms of psychopathology (SCL-90-R score) (β = 2.19, 95% CI = 0.24,4.14, p = 0.028) were significantly associated with symptom change at 6 months. Moreover, self-efficacy was associated with the outcome, however no longer significant in the multiple regression model. In multiple logistic regression analysis, anxiety severity at baseline (OR = 0.54, 95% CI = -1.13,-0.12, p = 0.018) was significantly associated with remission at 6 months. There was no predictive performance of the machine-learning models. Our study contributes with information that could be valuable knowledge for managing anxiety disorders in primary care.