Detailed study of the usage of the Rudjer Boskovic Institute's Library web site has been undertaken. Server log files are used to gain quantitative usage information. It is very important to understand how users access library collections, what problems they find, and what their usage patterns are. Compiled and interpreted properly, log information provides a baseline of statistics that indicate use levels. Such analysis also provides some technical information regarding server load, unusual activities, or unsuccessful requests, and can assist in marketing and web site development, as well as management activities. Data captured in the log files are discussed. The difficulties of log file analysis are presented. The limitations of log file data, Web log analysis software, and the inherent nature of the Web mean that log file statistics should be interpreted extremely cautiously.