Caching Web objects has become a common practice towards improving content delivery and users? servicing. A Web caching framework is characterized by its cache replacement policy, which identifies the objects (i.e. the elements on a Web page, which include text, graphics, and scripts) to be replaced in a cache upon a request arrival. In this paper, we present a cache replacement algorithm (so-called FRES-CAR), which identifies the objects that should be evicted by considering together three important criteria: object?s frequency, recency and size. Experimentation under synthetic workloads has shown that FRES-CAR achieves higher hit rates when compared with the most popular and existing algorithms.