Objective: The diagnostic value of the addition of alarm symptoms in distinguishing functional from organic gastrointestinal disease remains uncertain. We aimed to establish the value of alarm features in differentiating between organic disease and irritable bowel syndrome (IBS) and functional dyspepsia (FD). Methods: A total of 568 consecutive patients (63% female; mean age 44.7 years) completed a detailed symptom questionnaire and then received a complete diagnostic workup, as required. Questionnaire data were collected prospectively and audited retrospectively; the treating physician was blinded to the results of the questionnaires. Patients were coded and allocated to the following diagnostic groups: IBS, FD, organic diseases of the upper gastrointestinal tract, or organic diseases of the lower gastrointestinal tract. Logistic regression was used to identify the best subset of symptoms that discriminated organic disease from functional illness. Separate models compared IBS (n = 214) with diseases of the lower gastrointestinal tract (n = 66), and FD (n = 70) with diseases of the upper gastrointestinal tract (n = 250). Results: Age (50 years at symptom onset: odds ratio (OR) 2.65 (95% confidence interval 1.4-5.0); p = 0.002) and blood on the toilet paper (OR 2.7 (1.4-5. 1);p = 0.002) emerged as alarm features that discriminated IBS from lower gastrointestinal illness. A diagnosis of IBS was typically associated with female sex (OR2.5 (1.3-4.6); p = 0.004), pain on six or more occasions in the previous year (OR 5.0 (2.2-11.1); p<0.001), pain that radiated outside of the abdomen (OR 2.9 (1.4-6.3); p = 0.006), and pain associated with looser bowel motions (OR 2.1 (1.1-4.2); p = 0.03). A model incorporating three Manning criteria and alarm features yielded a correct diagnosis of IBS in 96% and a correct diagnosis of organic disease in 52% of cases. Alarm features did not discriminate FD from upper gastrointestinal disease. Patients with FD were significantly more likely to report upper abdominal pain (OR 3.7 (1.7-8.3); p = 0.002) and significantly less likely to report aspirin use (OR 0.26 (0.1-0.6); p = 0.001). The predictive value of symptoms in diagnosing FD was only 17%. Conclusions: Symptoms plus alarm features have a high predictive value for diagnosing IBS but the predictive value for a diagnosis of FD remains poor. Current criteria for the diagnosis of IBS should incorporate relevant alarm features to improve the diagnostic yield.