As the globe grows more interconnected, more malicious actors are taking advantage of network vulnerabilities. Network Intrusion Prevention Systems (NIPS) can be deployed to mitigate some of these risks; however, they can become a bottleneck for a quickly expanding network, since each packet from each network conversation must be examined. As this is a very memory-intensive task, this paper implements a new class of NIPS that uses an aggregated conversation summary called NetFlow, analyzed through a machine learning algorithm, to detect malicious activity while packets pass through at full speed. The proposed NIPS (NF-Based-NIPS) was then tested against a well-established NIPS known as Snort to determine its effectiveness, where it was able to keep up with Snort for the most part. Additionally, a comparative analysis of previous works was performed to determine which NetFlow features should be fed into the machine learning classifier.