Nonviral episomal vectors represent attractive alternatives to currently used virus-based expression systems. In the late 1990s, it was shown that a plasmid containing an expression cassette linked to a scaffold/matrix attached region (S/MAR) replicates as a low copy number episome in all cell lines tested, as well as primary cells, and can be used for the genetic modification of higher animals. Once established in the cell, the S/MAR vector replicates early during S-phase and, in the absence of selection, is stably retained in the cells for an unlimited period of time. This vector can therefore be regarded as a minimal model system for studying the epigenetic regulation of replication and functional nuclear architecture. In theory, this construct represents an almost "ideal" expression system for gene therapy. In practice, S/MAR-based vectors stably modify mammalian cells with efficiencies far below those of virus-based constructs. Consequently, they have not yet found application in gene therapy trials. Furthermore, S/MAR vector systems are not trivial to handle and several critical technical issues have to be considered when modifying these vectors for various applications.