Non-integrating gene-delivery platforms demonstrate promise as potentially ideal gene-therapy vector systems. Although several approaches are under development, there is little consensus as to what constitutes a true 'artificial' versus an 'engineered' human chromosome. Recent progress must be evaluated in light of significant technical challenges that remain before such vectors achieve clinical utility. Here, we examine the principal classes of non-integrating vectors, ranging from episomes to engineered mini-chromosomes to true human artificial chromosomes. We compare their potential as practical gene-transfer platforms and summarize recent advances towards eventual applications in gene therapy. Although chromosome-engineering technology has advanced considerably within recent years, difficulties in establishing composition of matter and effective vector delivery currently prevent artificial or engineered chromosomes being accepted as viable gene-delivery platforms.