Developing an OCR (Optical Character Recognition/ Reader) system requires expertise in various fields of researches, and the fact has been recognized as an obstacle to expanding the range of OCR applications, improving the recognition accuracy, and to providing end users with better OCR applications and services. This paper proposes a platform for distributed and cooperative OCR systems, called "OCRGrid," that allows end users to search for and use OCR servers over networks. We discuss the potential of the platform, useful applications, and problems that need to be addressed. This paper also introduces a toolkit for realizing Web-based OCRs and an implementation of the character recognition based on the majority logic.