The 2D bridge plans play an essential role in renovations or new replacements of transport infrastructures. As an indispensable component of plans, text serves as dimension labels, descriptions of materials, etc. Optical Character Recognition (OCR) can convert text in image into machine-encoded format, and has high potential and invaluable practical worth for analyzing, managing, and extracting textual information from plans. First, the selection of OCR-algorithms and the sampling of textual fragments from 2D bridge plans, including historic plans with handwritten text and digital plans with printed text, was carried out. Subsequently, the performances of different open-source OCR-algorithms on different textual fragments from 2D bridge plans are compared. In this process, the accuracy at character level and word level are chosen as two main factors to evaluate their performances. Finally, the most potential OCR-algorithms for the text recognition on 2D bridge plans are proposed.