Cloud computing is one of the most exciting paradigm shifts in distributed computing. Cloud computing is being widely used due to its readily available services at low cost. When the number of cloud users increase this may subsequently lead to data security and privacy threats. Data confidentiality and efficient data retrieval are major issues which block users to adopt cloud computing. The focus of this paper is data confidentiality and efficient data retrieval issues in cloud computing. The data classification and cloud model are proposed to overcome data confidentiality and efficient data retrieval issues. In this paper we propose a new cloud model, i.e., the Hybrid Multi-Cloud Data Security (HMCDS) model with data classification. The model is based on multi-cloud, different clusters and data classification. Two levels of security are considered, i.e., model and data classification based levels. © 2013 IEEE.