Computer programming has been gradually emphasized in recent computer literacy education and regarded as a requirement for all middle school students in some countries. To understand young students' perceptions about their own learning in computer programming, this study aimed to develop an instrument, Computer Programming Self-Efficacy Scale (CPSES), for all students above middle school levels. Based on Berland and Lee's computational thinking framework, this study developed the CPSES items at a literacy level and finally the instrument included the five subscales: "Logical Thinking," "Algorithm," "Debug," "Control," and "Cooperation." An exploratory factor analysis and reliability tests were conducted in this study. The reliability alpha was 0.96 for the overall scale, and ranged from 0.84 to 0.96 for the subscales. This study also confirmed the positive correlation between computer programming experience and computer programming self-efficacy. In addition, for low- and middle-experienced learners, significant gender differences were found in two subscales: Algorithm and Debug. The CPSES can be applied as an evaluation tool in computer education, robotics education, as well as integrated STEM or STEAM education in which computer programming was regarded as a part of computer literacy.