Controller area network (CAN) is a widely used fieldbus protocol in various industrial applications. To understand the network behavior under errors for the optimal design of networked control systems, the message response time of the CAN network needs to be analyzed. In this study, a novel delay time distribution analysis method for the response messages is proposed when considering errors. In this method the complex message queues are decomposed into typical message pattes and cases. First, a stochastic fault model is developed, and the probability factor is defined to calculate the error distribution. Then the message delay time distribution for the single slave node configuration is analyzed based on the error distribution. Next, based on the delay time distribution analysis of typical pattes and cases, an analysis framework of message delay time distribution for the master/slave configuration is developed. The testbed is constructed and case studies are conducted to demonstrate the proposed methodology under different network configurations. Experimental results show that the delay time distributions calculated by the proposed method agree well with the actual observations.