John Hawks has a post up, Multiregional evolution lives!, in response to Rex Dalton's reporting on Neandertal-human admixture. He notes:
These ongoing studies are concluding that present-day genetic variation is inconsistent with a simple model where a random-mating ancestral population gives rise to today's global population by means of a staged out-of-Africa dispersal. They next look at a model with some substantial (possibly complete) isolation between ancient human populations followed by a subsequent out-of-Africa dispersal. They show that this model fits the data significantly better. So far, so good. For a moment, I'm going to adopt a critical perspective. Previous results haven't yet been able to answer an important possible question: Can they distinguish the effects of intermixture outside Africa from an ancient population structure inside Africa? Increasingly it looks like population structure inside Africa may have been very important to the evolution of Late Pleistocene Africans. How can we distinguish these kinds of structure from each other? The short answer is that maybe we can't, yet. Human population history was not simple. If we take a simple model and add more parameters, it will fit the data better. The question is whether there may be some even better model with the same number of parameters. Population structure within Africa, selection on some loci but not others, asymmetrical migration -- all these and more might be possible.
The Out of Africa + total replacement model had a clean elegance, but it might not be viable in the near future. That being said it seems to me that the old Multiregional model implied, though proponents were often careful to reject this characterization, more regional parity than was the case. I do not expect the predominant African ancestry of modern humans to be rejected for example. There are other frameworks out there, such as Alan Templeton's Out of Africa again and again (Richard Dawkins favors this in The Ancestor's Tale).