The production of recombinant protein from mammalian cells is a key feature of the biotechnology industry. However, the generation of recombinant mammalian cell lines is still largely performed on an empirical basis and there are many potential areas for enhancement. We have shown previously that despite two rounds of limiting dilution cloning (LDC) of recombinant cell lines, there remained a high degree of heterogeneity in the resulting cell lines. We suggested that a rapid phenotypic drift occurred with these cells. It was unclear if this was a consequence of the added burden of production of a recombinant protein, the selection procedures, or merely an inherent feature of cell growth in culture. To address this, we have subjected untransfected (parental) cells to three successive rounds of LDC and monitored the growth properties of the resultant cells. The results show that despite repeated rounds of cloning, it was not possible to obtain phenotypically similar cell lines. We also demonstated that this phenotypic drift is not due to gross changes in the protein p27, a key regulators of the cell cycle. Although cells with a range of growth properties were observed even after three rounds of cloning, the variation in growth patterns between cell lines decreased after cloning. Hence, we suggest that by cloning it may be possible to generate untransfected cells, which have particular growth properties. Starting with a well-defined population of parental cells may aid in the subsequent generation of tranfectants with desired growth properties. © 2006 Wiley Periodicals, Inc.