idg,
BCARs used to demand that Vmcg was demonstrated in a 7 knot adverse crosswind, that being statistically the worst that you might reasonably expect to get on the rare occasion that an engine failed at the critical moment. As part of the process of harmonising national regulations to create JARs the CAA was reluctantly forced to accept zero crosswind Vmcg so now the US and European regs are the same in this respect. A lot of chicanery used to go on with some company,s test pilots loading the nosewheel to create more damping of the swing due to engine failure, but the certification authorities are pretty wise to this now and demand that the test is done with minimal weight on the nosewheel.
To gain a C of A the large aircraft manufacturer has to demonstrate crosswind TO and Landing in a crosswind of 0.2 Vs, except that he must demonstrate at least 20 knots but does not have to demonstrate more than 25 knots. Most manufacturers try to test the most they can, but the wind doesn't always blow when you want it and they cannot hold up the certification programme while they wait for it to do so. The max demonstrated crosswind is not a limit necessarily, but anything above that is to be treated with caution. Airlines will differ in their policy on this. Also, some manufacturers will quote a max crosswind tested in the simulator. However, this also needs to be treated with caution because no simulator is better than the data fed into it and it is doubtful whether, for example, the roll power due to aileron is acurately modelled for all the levels of ground effect which might be experienced in a crosswind landing. Personally I think that a hard limit should be set at the max crosswind demonstrate by the manufacturer, but I should think any proposal along those lines would result in some pretty loud squeals from industry.