With the evolution of World Wide Web many conveniences came to our way, but along with these facilities came some challenges like unlimited information resources and a large corpus of online data. Recommendation Systems have emerged as a solution to this information overload problem. They facilitate users by providing suggestions that effectively prune large information spaces so that users are directed toward items that best meet their needs and preferences. One specific domain is News Recommendation, where thousands of news sources are available online making it difficult for users to find an article relevant to their reading interests. This paper presents an algorithm that takes advantage of predefined categorization done by many online news sources, first to identify user interests and form a relevant user profile out of it and later to recommend news articles that might be of interest to the user. The proposed algorithm adopts collaborative filtering approach to selectively shortlist news articles that should be recommended to users. Outcome is a Personalized News Recommendation System that builds user profiles, requests feedback from users on recommended articles and uses feedback received to continuously monitor dynamically changing user reading patterns, keeping track of articles highly appreciated by users. The paper also presents results received on simulating proposed algorithm on a sample database of college students which indicate good satisfaction level for recommended articles.