Gene repair, the precise modification of the genome, offers a number of advantages over replacement gene therapy. In practice, gene targeting strategies are limited by the inefficiency of homologous recombination in mammalian cells. A number of strategies, including RNA-DNA oligonucleotides (RDOs) and short DNA fragments (SDFs), show promise in improving the efficiency of gene correction. We are using GFP as a reporter for gene repair in living cells. A single base substitution was introduced into GFP to create a nonsense mutation (STOP codon, W399X). RDOs and SDFs are used to repair this mutation episomally in transient transfections and restore green fluorescence. The correction efficiency is determined by FACS analysis. SDFs appear to correct GFP W399X in a number of different cell lines (COS7, A549, HT1080, HuH-7), although all at a similar low frequency (∼ 0.6% of transfected cells). RDOs correct only one of our cell lines significantly (HT1080-RAD51), these cells overexpress the human RAD51 gene; the bacterial RecA homologue. The GFP W399X reporter is a fusion gene with hygromycin (at the 5′ end), this has allowed us to make stable cell lines (A549, HT1080) to study genomic correction. Initial studies using our correction molecules show only low efficiencies of genomic repair (∽ 10− 4). Polyethylenimine (PEI) is used to deliver RDOs and SDFs into mammalian cells in culture for our study. We have used fluorescently labelled RDOs and SDFs to study the effectiveness of this process. FACS analysis of transfected nuclei implied efficient delivery (> 90%) both with SDFs and RDOs. However, confocal fluorescence microscopy suggests that a large proportion of the complexed RDO/SDF appears to remain outside the nucleus (or attached to the nuclear membrane). On the basis of these data we are assessing new delivery methods and factors that may alter recombination status to optimise gene repair.