There are probably a couple of factors here. Firstly, if they don't use an application server, their architecture could work wuite differently. Also, on live sites it is common to put a web server in front of the application server and use this to hide the application context. In effect it maps request to the web server domain to requests to the application server context. It should also be possible to create a servlet mapping of /*.pdf for your servlet rather than /download. You'll need to look at the API to find out where the filename turns up (I've used similar code several times, but I keep forgetting ) The last two points combined would allow you to use a servlet to control pdf downloads in the form "/myfile.pdf". I usually use the alternate servlet mapping to make a servlet act as a virtual directory, and then it manages all requests for resources in that directory. If you map the servlet as "/download/*", then you can lookat the file requested and decide what to do programatically before sending it back. Dave.