Double strand-break (DSB) induction allowed efficient gene targeting in barley (Hordeum vulgare), but little is known about efficiencies in its absence. To obtain such data, an assay system based on the acetolactate synthase (ALS) gene was established, a target gene which had been used previously in rice and Arabidopsis thaliana. Expression of recombinases RAD51 and RAD54 had been shown to improve gene targeting in A. thaliana and positive-negative (P-N) selection allows the routine production of targeted mutants without DSB induction in rice. We implemented these approaches in barley and analysed gene targeting with the ALS gene in wild type and RAD51 and RAD54 transgenic lines. In addition, P-N selection was tested. In contrast to the high gene targeting efficiencies obtained in the absence of DSB induction in A. thaliana or rice, not one single gene targeting event was obtained in barley. These data suggest that gene targeting efficiencies are very low in barley and can substantially differ between different plants, even at the same target locus. They also suggest that the amount of labour and time would become unreasonably high to use these methods as a tool in routine applications. This is particularly true since DSB induction offers efficient alternatives. Barley, unlike rice and A. thaliana has a large, complex genome, suggesting that genome size or complexity could be the reason for the low efficiencies. We discuss to what extent transformation methods, genome size or genome complexity could contribute to the striking differences in the gene targeting efficiencies between barley, rice and A. thaliana.