Previously described transmission studies have shown that HIV strains isolated from individuals infected from a common source are more homogeneous than HIV strains isolated from individuals with unrelated infections. This has been the basis, in at least four instances, for deciding whether apparently epidemiologically related cases represent actual transmissions. To date, HIV transmission studies have usually included sequence data from the most likely source of infection, and the probability of transmission from the donor to the recipient has been assessed by measuring sequence similarity against control data using likelihood analysis. We have recently studied a putative transmission involving a UK health care worker (CPHL1), a patient of CPHL1 (CPHL2), and CPHL3, a member of the same "sex circle" as CPHL2. We have used sequence distance and neighbour joining methods as well as likelihood analysis as means of determining genetic relatedness. Though no other source of infection was available our findings did not support the possibility that CPHL1 had infected CPHL2. Strain CPHL3 was closer to CPHL2 than to CPHL1. It is shown that control data from documented transmission events can be used to establish the source of infection in the absence of an index case. It is also shown that the C2-V3 region analysed in previous transmission studies is unreliable for accurate phylogenetic analysis. The results indicated that gp120 is a more informative region than C2-V3 for molecular transmission studies. Sequence distances between the env genes of related and unrelated infections have been derived in this work.