The LD and 'other' votes are high, but they won't necessarily last - both are benefitting from recent publicity from the local, European and 2 by-elections we had in the last couple of months. There is definitely an idea of 'momentum' in British politics - plenty of people only consider voting for a party when they see enough other people voting for it that it doesn't look so ridiculous.
There's a site which attempts to track UK opinion polls, and translate them into seats in parliament -
http://www.financialcalculus.co.uk/election/index.html .
If you click on 'opinion polls', you get a history of the main 3 parties' polls over the last 18 months or so. There was a point about September 2003 when all three parties were on about the same vote (probably when the Hutton Inquiry was starting to hear evidence), but it didn't last.
If you click on 'battlemap', you can translate poll numbers into predicted seats (it's not a sophisticated formula; the details are on the site). Note that 'power-use' says you need to adjust for the 'other' vote; eg for this '30,28,28' poll, adjust the Labour and Conservative numbers as 30 * 93.6 / (30+28+28) and 28 * 93.6 / (30+28+28) respectively, giving roughly 32 and 30, before using the grid. This gives Labour 349, Tory 174 and Lib Dem 92 - still a Labour majority in Parliament.
So I don't think Labour will panic yet. It's only if they start to reguarly poll less than the Tories that they might fall apart - if that meant a fight between Blair and Brown, or others, anything could happen. But looking at that website grid, it looks as though Labour need an (adjusted) vote of 30% or below (say 28% in real figures), or to trail the Tories by a point, just to lose their majority. They'll still be the largest party in parliament almost whatever happens.