This document presents a method for credit card fraud detection using Hidden Markov Models. It discusses how HMMs can be trained with normal cardholder behavior and then used to determine if incoming credit card transactions have a sufficiently high probability of matching that normal behavior. Transactions that do not match are considered fraudulent. The method aims to identify fraudulent transactions while avoiding rejection of genuine ones. The document also compares HMMs to other fraud detection methods and explains why HMMs are preferred.