The plastid genome (ptDNA) of higher plants is highly polyploid, and the 1000–10 000 copies are compartmentalized with up to approximately 100 plastids per cell. The problem we address here is whether or not a newly arising genome can be established in a developing tobacco shoot, and be transmitted to the seed progeny. We tested this by generating two unequal ptDNA populations in a cultured tobacco cell. The parental tobacco plants in this study have an aurea (yellowish–golden) leaf color caused by the presence of a barau gene in the ptDNA. In addition, the ptDNA carries an aadA gene flanked with the phiC31 phage site-specific recombinase (Int) attP/attB target sites. The genetically distinct ptDNA copies were obtained by Int, which either excised only the aadA marker gene (i.e. did not affect the aurea phenotype) or triggered the deletion of both the aadA and barau transgenes, and thereby restored the green color. The ptDNA determining green plastids represented only a small fraction of the population and was not seen in a transient excision assay, and yet three out of the 53 regenerated shoots carried green plastids in all developmental layers. The remaining 49 Int-expressing plants had either exclusively aurea (24) or variegated (25) leaves with aurea and green sectors. The formation of homoplastomic green shoots with the minor green ptDNA in all developmental layers suggests that the ptDNA population in a regenerating shoot apical meristem derives from a small number of copies selected through a stochastic process.