Information retrieval functions in digital libraries have a different context from those in searching commercial databases or OPACs. After defining digital libraries focussing on their types, this paper describes the characteristics of browsing methods from several aspects. The method is indispensable as a means of accessing the needed information. Because digital libraries should be user-oriented, browsing should reflect the information-seeking behavior of human beings. The retrieval functions, however, should not be limited to browsing but ordinary Boolean search methods are also necessary. Printed materials and microfilms are converted into electronic form using OCR in the digital library environment. It is inevitable that digital documents produced by scanning and OCR contain recognition errors. As a result, if keywords contained in digital documents are misspelled, they will cause mismatching during a search. This paper also deals with the approximate matching principle based on n-gram, which is supposed to solve such problems.