The enormous diversity of the human immunodeficiency virus (HIV) has led to the idea that designing vaccines to specific geographic regions, or clades, could simplify the complexity of the task. Yet, despite the sequence diversity, all HIV viruses known to date interact with the same cellular receptors (CD4 and/or a coreceptor, CCR5 or CXCR4). In this review we examine the existing evidence to support a clade-specific vaccine strategy for induction of neutralising antibodies. We concentrate on lessons learnt from natural infection of humans. In short, the vast majority of studies to date indicate that neutralisation of HIV-1 is not clade specific. Potent sera tend to neutralise a range of heterologous viruses with no apparent clade preference, and none of the human neutralising monoclonal antibodies so far generated demonstrate significant clade preference. All but one of the most broadly neutralising antibodies are to functional regions involved in receptor interactions and plasma membrane fusion. Given these facts, we suggest that vaccine approaches that focus on 'clade-specific' and 'clade-generic' vaccines will logically converge on the same functionally conserved envelope structures. It still remains to be determined whether or not the task of designing a 'clade-generic' vaccine could be simplified by focusing on the viral envelopes with 'transmitting phenotypes'.