Methods and apparatus for selectively caching (and de-caching) video content in network so as to reduce content transformation requirements and also cache storage requirements. In one embodiment, a content caching controller associated with a content server differentiates content requests based on content attributes such as the requested codec format (e.g., MPEG or Windows Media), resolution, bitrate, and/or encryption type or security environment. If the content requested by a user is not available with the requested attribute(s), the content server transfers to content to the user by first transforming it. The content server also speculatively caches the transformed content locally, so that a future request for the same content with the same attributes can be filled by transferring without the intermediate transformation step. The controller allows the network operator to optimize use of available storage and transcoding resources.