The consumption of media has changed with increased penetration of broadband and mobile networks. Today, media streaming is used broadly in various applications and is also being widely used to create new business models such as Video or Media on Demand. Due to the real-time requirements in streaming, overload prevention is vital to guarantee an uninterrupted play of the media. Unfortunately, client demands for media services are highly variable and required server capacities are therefore hard to predict. Admission control is used to accept or deny incoming service requests in order to avoid overload, but existing media streaming software includes only limited support for admission control by allowing for pre-defined static rules. While in media streaming the bottleneck resource (CPU, disk I/O, etc.) might change over time, such rules define thresholds concerning the utilization of only a single resource: the bandwidth. This work introduces a method for adaptive admission control in the presence of multiple scarce resources, which allows predictive rejections of service requests to reserve capacity for future requests with higher revenues in order to maximize overall profit.