This chapter presents several main approaches to the design of information granules. It deals with the principle of justifiable granularity to show how a single information granule is constructed in the presence of available experimental evidence. The chapter explores methods of clustering, especially fuzzy clustering, as a vehicle to construct information granules. It looks at the augmentation of clustering techniques so that they result in information granules built on the basis of numeric data as well as knowledge tidbits of human expertise which are some information granules themselves. The chapter also presents the idea of collaborative clustering in which higher-type information granules are formed on the basis of several sources of data. An operational framework depends upon the accepted formalism of information granulation, namely, a way in which information granules are described as sets, fuzzy sets, shadowed sets, rough sets, probabilistic granules, and others.