Open government data has become an important movement for government administrations around the world. Nevertheless, government agencies tend not to open their datasets and can act reluctantly to embed the concept of data publication in their daily operations. Accordingly, this study examines the determinants that are argued to have influences on the government agencies' intention and behavior of open data publication. Relevant hypotheses are developed through literature review to form a preliminary research model, and respective influences of the determinants on the government agencies' intention and behavior to publish datasets are statistically tested. The empirical results show that the determinants account for 60.4% of the variance in agencies' intention and 54.2% of the variance in agencies' behavior. All the determinants are statistically significant except for perceived effort and perceived benefits. Specifically, facilitating conditions and organizational capability are the two factors having the strongest positive effect. Perceived usefulness, external influence, and organizational culture also have positive influences while perceived risks indeed incur a negative impact on the intention. On the contrary, perceived effort and perceived benefits are found to be insignificant. Lastly, intention is statistically significant and indeed acts as a strong antecedent to predict government agencies' activities of open data publication. Practical implications are also offered to practitioners. A designated government agency should act as the coordinator to help agencies obtain facilitations from others. Agencies' concern of data misuse should be mitigated, as it is found to be the major driver having a negative impact. The limitations and future research directions are also discussed. The findings and discussions of this study are expected to contribute to the current open data literature. •The influential factors of open data publication are examined and discussed.•Facilitating condition and organizational capability are the two major drivers.•Perceived usefulness, external influence, and culture also have positive influences.•Perceived risks incur a negative impact while data misuse weighs the most.•Perceived effort and perceived benefits are not found to be significant in the study.