If, like me, you find polling interesting you should take a look at the fascinating statistical analysis being done by Nate Silver on the UK election.
Nate is credited with having accurately predicted 49 out of 50 states in the USA in the last election and has an excellent reputation. Whether he's able to pick up the nuances of the UK system, who knows, but the model seems to make sense to me.
Rather than working on universal swing Silver's model works on the shifting pattern of voters between all three parties, based on their 2005 results.
None of his modelling suggests that there's likely to be a majority Government, although the Conservatives could pick up just over 300 seats in some of his more optimistic modelling.
I'd love to make a prediction myself but at this point I think in all honesty, virtually nothing is certain. The polling is interesting, but as always, speculative, and unsupported by qualitative research to demonstrate is veracity.
One point which I will make in parting, although everyone is excited about May 6th, in reality its what happens on May 7th that matters for the country.