In this article, we propose an original way to characterize information content in Online Signatures through a client-entropy measure based on local density estimation by a Hidden Markov Model. We show that this measure can be used to categorize signatures in visually coherent classes that can be related to complexity and variability criteria. Besides, the generated categories are coherent across four different databases: BIOMET, MCYT-100, BioSecure data subsets DS2 and DS3. This measure allows a comparison of databases in terms of clients’ signatures according to their information content