Vectors derived from human immunodeficiency virus type 1 (HIV-1) are an attractive option for many gene therapy applications as they can transduce non-cycling cell populations, and can integrate their genome into the host cell chromosome. The rationale underlying the design of most retroviral vector systems is to segregate the viral cis sequences, which are required for transfer of the viral genome, from the trans sequences that encode viral proteins. This allows the efficient production of replication incompetent virus and has been successfully applied to the generation of HIV-1 vectors. Nonetheless, the possibility that recombination events in the vector production system can generate replication-competent virus, combined with the pathogenic nature of HIV-1, raises major bio-safety issues. Numerous HIV-1 vectors have now been reported, with each generation significantly improved in ways designed to reduce the risk of replication-competent virus being produced. However, progress in vector design needs to be complemented by the development of methods for the quantitation of the probability of replication competent virus being produced. Assaying individual events in the multi-step pathway that can lead to the production of replication-competent virus, rather than relying on the detection of replication-competent virus per se, will be important for quality control purposes. This review will specifically examine the approaches to HIV-1 vector design that have been postulated as increasing bio-safety, possible methods for evaluating bio-safety and whether these approaches are likely to be sufficient to overcome resistance to the use of HIV-1 for clinical application. In addition, we discuss the possible justifications for developing vectors from lentiviruses other than HIV-1.