A number of recent studies have identified classroom blogging as a promising approach to teaching writing in English as a second language (ESL) in non-Anglophone higher education settings. However, many of these research designs are methodologically inadequate and their results suffer from overestimated effects on students' learning performance. In this regard, whether or not classroom blogging can be safely recommended as an effective approach to teaching writing in the ESL classroom, as suggested by previous researchers, needs a well-designed empirical reassessment. For this very purpose, this paper reports on a study which attempts to overcome these limitations by conducting a 36-week, comparative experiment recruiting two groups of ESL undergraduates. The experimental group (EG) was taught how to write in English using a blog-based approach, while the control group (CG) was taught using traditional paper-based materials and methods. The writing performance, learning motivations, and self-efficacy beliefs of both groups were assessed using two sets of pre- and post-tests in terms of writing tasks and a questionnaire. The data collected were analyzed using paired-sample "t" tests, analysis of covariance, and multivariate analysis of covariance. The results of the study show that (1) both groups made significant improvements in all of their performance in writing and in their learning motivations and self-efficacy beliefs; (2) no significant difference was found between the two groups in terms of their writing proficiency; but a closer examination shows that (3) the CG significantly outperformed the EG in terms of learning motivation.