The objective of this study is to clarify factors which affect use of public libraries, using multiple regression models. Other studies in this area do not take the time factor into account. Therefore, this study compares five multiple regression models on five different points in time between 1965 and 1985. The criterion variable is per capita circulation, while explanatory variables are data on library resources (per capita collection, etc.) and community profile. All twenty-three Tokyo wards and all of the cities in Tokyo metropolitan are selected for analysis. Results indicate that the effects of explanatory variables on per capita circulation are different at different points in time. The variables which explain the variance in per capita circulation at all the points in time are: (1) per capita collection, (2) population of community per thousand. librarians, (3) daytime population, and (4) proportion of professional or administrative workers in the working population. Analysis of correlations between all explanatory variables shows that behind these four determinant variables are the following three fundamental factors of library use: (1) resourceful of the library, as measured by, for example, per capita collection and population per thousand librarians, (2) whether the library is located in a commercial or residential area, and (3) education. The second point shows that public libraries in Tokyo have adjusted themselves mainly to the needs of residential areas, rather than those of commercial areas.
