The invention discloses an allocation method and device for video monitoring tasks and relates to the field of cloud computing. The method comprises the following steps of selecting history tasks having higher resource utilization rate than a preset value from history video monitoring tasks as a training sample set; determining attribute features of the current video monitoring task; determining K history tasks which are closest to the attribute features of the current video monitoring task from the training sample set; predicting resource allocation needed by the current video monitoring task according to the resource allocation of the K history tasks; and allocating a virtual machine to execute the current video monitoring task according to the predicted resource allocation needed by the current video monitoring task. According to the method and the device, the resource allocation needed by the current video monitoring task is predicted based on the resource allocation of the history video monitoring tasks with the similar attribute features, so that system resources are reasonably allocated and the resource utilization rate is improved.