Background As there is no biological marker for irritable bowel syndrome, a diagnosis is made using symptom-based criteria. Aim To evaluate the stability of self-reported symptoms consistent with Rome II-based irritable bowel syndrome classification. Methods Irritable bowel syndrome subjects identified in a 2001 population-based study by modified Rome 11 criteria were re-contacted 2 years later. Data were collected via a web-based questionnaire. Results Of the 697 subjects, 30% remained in the same irritable bowel syndrome subtype in both surveys, 18.4% changed irritable bowel syndrome subtype and 52% no longer met the irritable bowel syndrome criteria at follow-up. Subjects continuing to meet the irritable bowel syndrome criteria were more likely to have been initially classified in the alternating irritable bowel syndrome subtype and had more psychological impairment and lower irritable bowel syndrome-related quality of life than subjects not fulfilling the irritable bowel syndrome criteria at follow-up. Lack of pain caused more subjects to fall out of the irritable bowel syndrome criteria than the absence of non-painful bowel symptoms. However, the majority of subjects that did not fulfill the pain component of the irritable bowel syndrome criteria continued to report abdominal pain of at least moderate severity. Conclusion In a US population-based follow-up study using modified Rome II criteria, we found irritable bowel syndrome is episodic in nature and current classification is limited in capturing fluctuation of disease over time.