From the New Yorker 1/10/2011: http://www.newyorker.com/reporting/2011/01/10/110110fa_fact_peed
The banana variety known as the Cavendish makes up 97% of exports and is pretty much the only one consumed in the US and Europe. Until the 60s, the hardier, sweeter and larger Gros Michel was the strain of choice until it was wiped out worldwide by a soil-borne fungus. The Cavendish is currently undergoing the same process; it has been wiped out in Asia and it is only a matter of time before the fungus spreads to South America.
There are currently no viable alternatives — the thousands of other strains available worldwide are some combination of worse tasting, smaller (more expensive to grow), or less hardy (more difficult to transport and maintain freshness). We have downgraded our banana-eating in the past and it is likely we will be doing so again shortly.
There are two efforts to breed a fungus resistant Cavendish at the moment. In Australia (and likely there are many similar efforts), there are attempts to splice different varieties of genes from other plants and animals into the Cavendish genome which could combat the effects of the fungus. Since GMO food is frowned upon in the US and Europe, any fruits of this labor are likely not going to be commercializable there and our hopes are currently pinned on the painstakingly slow process of trying to cross-breed different varieties one generation at a time (each cycle takes several years) by South American plant breeders.
In practice, there is no difference between breeding plants/animals and GMO other than GMO allows faster and more precise outcomes. Could we create some kind of crazy carcinogenic mutant food? Yes, but that can be created just as easily through regular breeding. Due to the speed of GMO strain creation, we do however increase the probability of such outcomes. I think there needs to be a utility calculation performed to decide on a case-by-case basis whether GMO is the optimal approach for different foods/target markets. Engineering vitamin and mineral-rich rice for Africa is probably a much better idea than trying to make the bananas a little tastier for countries that already have access to a wide variety of dietary substitutes.