A method and system for detecting and preventing Internet fraud in online transactions by utilizing and analyzing a number of parameters to uniquely identify a computer user and potential fraudulent transaction through predictive modeling. The method and system uses a delta of time between the clock of the computer used by the actual fraudulent use and the potentially fraudulent user and the clock of the server computer in conjunction with personal information and/or non-personal information, preferably the Browser ID.