The ERIC network conversion to an OCR (Optical Character Recognition) mode of data entry was studied to analyze the potential effectiveness of OCR data entry for future EPC/s (Editorial Processing Centers). Study results are also applicable to any other system involving decentralized bibliographic data capture and conversion functions. The report does not recommend such OCR applications unless: (1) all OCR keying can be done on standardized forms; (2) consistent high quality of character images keyed on OCR forms can be maintained; and (3) suitable methods can be devised for coping with special notation (e.g., mathematical, chemical). The report contains performance data on both the OCR keying and OCR scanning functions, and provides a cost/savings analysis. Practical guidelines are provided for the implementation and operation of OCR data entry systems. (Author)