Autonomously replicating virus-based vectors have been investigated as a means of introducing heterologous genes into plants. This approach has a number of potential advantages over stable genetic transformation, particularly in terms of speed and levels of expression that can be obtained. Several groups of plant viruses, with genomes consisting of both DNA and RNA, have been investigated as possible gene vectors. In the case of DNA viruses, it has generally been possible to identify nonessential regions of the genome that can be replaced by foreign sequences. However, there appear to be limitations on the size of insert which can be tolerated. In the case of RNA viruses, replacement of viral sequences usually has a drastic effect on the viability. However, in several cases it has proved possible to substantially increase the size of the viral genome by the direct insertion of additional sequences while still retaining the ability of the viruses to multiply and spread in plants. These RNA virus-based systems appear to have the greatest potential as gene vectors.