Purpose – Increased awareness regarding environmental issues has encouraged corporations to disclose carbon related information. The study concerns carbon disclosure in the typical form of the Carbon Disclosure Project (CDP). The CDP constitutes an international voluntary code developed by a non-government organization to encourage consideration of carbon emission issues in decision making. Such a code can flexibly bridge the gap between individual companies’ sustainability initiatives and mandatory, legal regulation. We examine the effectiveness of the code and the determinants of the transparency of the carbon disclosure. The study contributes to the understanding of management (dis)incentives to be (un)accountable for their carbon impact on global warming.Design/methodology/approach – The paper is based on the academic literature on motivations for sustainability and environmental reporting, together with an analysis of the carbon disclosures made by the Global 500 firms. We assess the degree of transparency using the Carbon Disclosure Transparency Score (CDTS) adopted from the CDP reports. The CDTS focuses on carbon risks and opportunities, carbon footprints, carbon reduction activities, governance and verification of carbon information. We consider the influence of information needs of stakeholders and institutional effects on carbon transparency.Findings – We find the CDP has attracted an overwhelming majority of the Global 500 to disclose carbon information. However, we observe a great disparity in carbon transparency between firms in different sectors and institutions. Our results show 74% of 243 firms in our sample achieved 50% or higher on the CDTS (sample average is 60%), so their reports are reasonably transparent, and the remaining were un-transparent (or opaque). We interpret the lack of transparency as evidence of a lack of managerial incentive to be accountable. In addition, we find firm size, leverage, industry membership, emission trading scheme (hereafter ETS), stringency of environmental regulation, as predicted, are significantly associated with carbon transparency. Research limitations/implications –The results imply that stakeholder theory and institutional theory provide different but complementary explanations for the development of carbon disclosure practices. Future research is needed to gauge how the CDP code impacts front-line decision making. It could be useful to have further standardization of CDP reporting formats and contents, including more details about project-level implementation. Practical implications – The CDP in its current incarnation is a system which produces assessable reports of carbon activities, and it encourages the development of web-based forms of corporate accountability. From the organization’s perspective, the study provides immediate feedback and has the potential to improve practice through the identification of forces for change not considered in prior theorizing. Based on the findings, we make some suggestions for improvement of carbon disclosure. Some firms have already been involved in voluntary external assurance, which added creditability and value to the reports. Use of voluntary external assurance should be encouraged.Originality/value – Our study attempts to propose a research instrument to measure carbon transparency (obscurity) as a multi-item construct. Our findings help to illuminate the effectiveness of a voluntary code in the social construction of how stakeholder theory and institutional theory are to be interpreted and implemented. In addition, our international cross-sectional evidence for determinants of carbon transparency contrasts with findings in previous research focusing on general environmental issues, and research in national settings.