We present a combination of techniques for interactive out-of-core visualisation of isosurfaces from large time-dependent data sets. We make use of an index tree, computed in a pre-processing stage, which effectively captures temporal coherence in the data set. This tree data structure enables fast extraction of all isovalue-spanning cells from any time step and for any isovalue. For very large time-dependent data sets, such as those resulting from CFD simulations, this data structure can easily become too large to fit in main memory. Therefore, we have adapted the generation of the data structure, as well as the data structure itself for out-of-core application. During generation, the data set is spatially divided into several regions, each resulting in a separate tree. For visualisation, the application uses all these trees simultaneously, but will use only part of each of the trees. Only a user-specified time window will be kept in main memory and other parts of the tree will be read and released on demand. Finally, to avoid time-consuming triangulation and surface reconstruction, we have used a hardware-assisted direct point rendering algorithm for displaying the isosurfaces. These combined techniques allow interactive exploration and visualisation of very large time-varying data sets on a normal PC. Keywords: Scientific visualization; Large data handling; Isosurfaces; Out-of-core techniques; Time-dependent data@en