Under physiological conditions, many proteins that include a region lacking well-defined three-dimensional structures have been identified, especially in eukaryotes. These regions often play an important biological cellular role, although they cannot form a stable structure. Therefore, they are biologically remarkable phenomena. From an industrial perspective, they can provide useful information for determining three-dimensional structures or designing drugs. For these reasons, disordered regions have attracted a great deal of attention in recent years. Their accurate prediction is therefore anticipated to provide annotations that are useful for wide range of applications. POODLE-I (where "I" stands for integration) is a web-based disordered region prediction system. POODLE-I integrates prediction results obtained from three kinds of disordered region predictors (POODLEs) developed from the viewpoint that the characteristics of disordered regions change according to their length. Furthermore, POODLE-I combines that information with predicted structural information by application of a workflow approach. When compared with server teams that showed best performance in CASP8, POODLE-I ranked among the top and exhibited the highest performance in predicting unfolded proteins. POODLE-I is an efficient tool for detecting disordered regions in proteins solely from the amino acid sequence. The application is freely available at http://mbs.cbrc.jp/poodle/poodle-i.html.