A trust-based intrusion detection scheme for hybrid cloud computing is proposed. We consider a trust metric based on honesty, cooperation and efficiency for detecting malicious machines. We use Perron-Frobenius theorem to detect intrusion based on trust and observations. By statistically analyzing peer-to-peer trust distributed results, we apply trust-based intrusion detection to assess the trustworthiness and maliciousness. An analytical model and simulation for performance are developed. We analyze the sensitivity of false alarms with respect to the minimum trust threshold below where a node is considered malicious. Results confirm that our proposal is flexible enough to detect malicious behaviours in various context of executing application in hybrid cloud. With this work, we can guide future execution in the cloud resource.