Background For comparing trials using different classifications for irritable bowel syndrome (IBS) subtypes, it is important to know whether these identify the same sub-populations. Our aim was to determine the agreement between Rome II and Rome III subtypes, and to explore whether agreement depends on the symptom reporting method. Methods Rome II IBS patients from two identical, randomized placebo-controlled trials of probiotics were included. Retrospective subtypes were based on the Rome II questionnaire. Prospective subtypes were based on diary cards for 2 weeks of run-in. Agreement was determined between: (i) retrospective Rome II and Rome III, (ii) prospective Rome II and Rome III, (iii) retrospective Rome II and prospectively Rome III, (iv) retrospective and prospective Rome II, and (v) retrospective and prospective Rome III. Key Results A total of 126 patients, 72% women, mean age 46±15years, were included. The agreement between subtypes using the same symptom reporting method was: (i) 90.3% (κ=0.85) for retrospective subtypes, and (ii) 84% (κ=0.76) for prospective subtypes. The agreement between subtypes using different symptom reporting methods was, (iii) 49% (κ=0.23) for retrospective Rome II and prospective Rome III, (iv) 51% (κ=0.26) for Rome II subtypes, and (v) 41% (κ=0.25) for Rome III subtypes. Conclusions & Inferences Agreement between Rome II and Rome III subtypes is good to very good when using the same symptom reporting method. When mixing methods, agreement is only fair even within the same classification. This has implications for comparison of trials using different symptom reporting methods for subtyping.