In this paper we present several advances for managing caching resources in multimedia streaming systems. The improvements are related to increase the performance of these resources. First, we use an analytical model to build a caching algorithm for streaming known streams. This algorithm is optimum if the bandwidth requirements are constant over time. Then, we present two different caching algorithms that improve performance when the stream bandwidth requirements are variable. The adaptive interval caching algorithm (AIC) improves performance by dynamically selecting which streams should be cached depending on the variations of the bandwidth requirements. The adaptive and smoothing interval caching algorithm (ASIC) accomplishes the same goal and also includes smoothing capabilities that keep the storage bandwidth requirements more stable. The paper also proposes some modifications to increase global performance when the streams come from several sources. The evaluation shows that AIC and ASIC algorithms perform better than classical IC algorithms when used with variable bandwidth streams.