The production of Green Fluorescent Protein in recombinant NIH3T3 mouse fibroblast cells was used as a model to determine the optimal conditions for the rapid isolation of high-producing cell lines with a fluorescence-activated cell sorter. "Bulk sorting", that is, sorting of a large number of positive cells, did not result in a stable, high-producing cell line due to overgrowth of high-producing cells by low- or nonproducing cells. The production kinetics and expression of GFP during batch culture was found to differ between NIH3T3 cells and HepG2 hepatoma cells, even though the same plasmid was used for transfection. The kinetics of product formation need therefore to be determined from case to case to select the optimal timepoint for analysis and sorting. Subcloning of sorted cells into microtiter plates only resulted in high-producing subclones when 1 or 2 cells were seeded per well. Higher seeding rates again resulted in overgrowth of low- or nonproducers. By subcloning, two high-producing cells lines could be isolated. They had a 10- and 15-fold higher fluorescent signal compared to the negative control. While one of these subclones started to decrease it＇s GFP expression after 2 months, the other clone stably expressed GFP for 4 months.