In cloud computing, resources are provisioned and admitted on demand. During the runtime, workload can be calculated by using the workload analyzer and the resources are allocated. Deciding the resources needed is a difficult task in cloud computing. Autonomic provisioning of resources is needed to satisfy the future demands. In this paper, we present a Hybrid Admission Control and Resource Provisioning (HACRP) framework based on autonomic computing and Q-learning. We evaluate the performance of the framework using the Mediawiki traces. The proposed framework calculates the resources needed and allocates resources based on the demands. © 2019 NSP Natural Sciences Publishing Cor.