Can user-perceived latencies in the World Wide Web be improved by pre-loading a cache with resources that are predicted to be accessed in the future by the user? If so, how? And how can we measure the effectiveness of a system that peforms such Web cache pre-loading? This talk will provide some answers to these questions, focusing on the design and development of two different tools for evaluating both pre-loading and non-pre-loading Web-caching systems. For early evaluation in the development process, one tool simulates expected performance. The other tests fully implemented systems in a real-world environment. No previous knowledge of Web caching will be assumed.