A method for adaptive content balancing for Web clients includes monitoring content processing performance in different client computing devices communicatively coupled over a network to a content server, measuring the content processing performance for each of the different devices and storing the measured performance for each of the different devices in connection with different resource groupings of the requested content type. Thereafter, a request for content is received in the content server from one of the devices, and the stored performance is retrieved. As such, the retrieved metrics of the computing device are matched to the retrieved metrics of one of a multiplicity of groupings of one or more different resources associated with the requested content and the resources of the one of the multiplicity of the groupings is included in the requested content. Finally, the requested content is transmitted to the one of the client computing devices.