This is the simplest to set up and fastest running of the TM4J back-ends. Topic map objects created by the parser or by the application are just held in memory. Indexes are managed using the standard J2SE Collections.

The in-memory backend can easily accomodate topic maps of the order of low tens of thousands of objects in size with a JVM heap of 256MB, increasing the amount of memory available to the JVM also increases the capacity of the back-end.

The downside of the in-memory backend is that the data is not persistently stored. It is possible to simulate a form of persistance by using the import/export facilities of the Topic Map Engine to read from/write to XTM syntax files. This backend is also useful for web applications where the overall size of the topic map is not too large and the need for updates is minimal.