Background Anxiety disorders frequently recur in clinical populations, but the risk of recurrence of anxiety disorders is largely unknown in the general population. In this study, recurrence of anxiety and its predictors were studied in a large cohort of the adult general population. Methods Baseline, 3-year and 6-year follow-up data were derived from the Netherlands Mental Health Survey and Incidence Study-2 (NEMESIS-2). Respondents (N = 468) who had been in remission for at least a year prior to baseline were included. Recurrence was assessed at 3 and 6 years after baseline, using the Composite International Diagnostic Interview version 3.0. Cumulative recurrence rates were estimated using the number of years since remission of the last anxiety disorder. Furthermore, Cox regression analyses were conducted to investigate predictors of recurrence, using a broad range of putative predictors. Results The estimated cumulative recurrence rate was 2.1% at 1 year, 6.6% at 5 years, 10.6% at 10 years, and 16.2% at 20 years. Univariate regression analyses predicted a shorter time to recurrence for several variables, of which younger age at interview, parental psychopathology, neuroticism and a current depressive disorder remained significant in the, age and gender-Adjusted, multivariable regression analysis. Conclusions Recurrence of anxiety disorders in the general population is common and the risk of recurrence extends over a lengthy period of time. In clinical practice, alertness to recurrence, monitoring of symptoms, and quick access to health care in case of recurrence are needed.