Presented in this paper a scalable bloom filter based prefilter and a hardware-oriented predispatcher pattern matching mechanism for content filtering applications, which are scalable in terms of speed, the number of patterns and the pattern length. Prefilter algorithm is based on a memory efficient multi-hashing data structure called bloom filter. According to the statistics of simulations, the filter ratio can reach up to 60% if the whole engine has been trained well. It has been showed that this engine could enhance the capabilities of general-purpose IDS solutions.