It is uncertain whether the treatment will work in humans, but the San Francisco pharmaceutical company behind the research wants to test it on people soon.
One leader in the field of Alzheimer's disease research calls the vaccine a breakthrough.
"I think, even if it doesn't turn out to be effective in humans, it is going to open an area of research that we haven't had before and tell us much more about the disease," Bill Thies, vice president for medical and scientific affairs for the Alzheimer's Association, told CBS News.
Deposits in the brain of a sticky protein called amyloid are one of the characteristics of Alzheimer's. The vaccine appears to prevent the formation of these so-called plaques in mice that were genetically engineered to overproduce amyloid.
In their study, a team of researchers, led by Dale Schenk at Elan Corp., tried to trick the immune system of the mice to recognize amyloid as a foreign substance that should be attacked.
The researchers injected nine 6-week-old mice with amyloid combined with substances that excite the immune system. Seventeen other mice of the same age did not get the vaccine.
When the mouse brains were dissected after a year, the researchers were surprised to find no or very small plaques in the injected mice; the unvaccinated mice had extensive deposits.
The researchers then tried a more ambitious experiment: injecting the vaccine into 24 one-year-old mice that already had plaques. Twenty-four similar mice did not get the vaccine.
"We saw that it completely stopped the further progression of the disease," Schenk told Reuters. "It looks like it might have actually diminished the plaques."
Elan wants to start trials with people later this year. Schenk said researchers hope to submit a vaccine to the Food and Drug Administration for approval in five years.
But there are a number of reasons the method may not prevent or halt Alzheimer's in humans, particularly if the amyloid plaques may be a symptom of the disease, rather than the cause.
Also, Alzheimer's patients have other changes in the brain that the mice do not fully exhibit, such as tangles of protein inside nerve cells, said Dr. Blas Frangione, head of the Alzheimer's research unit at the New York University School of Medicine.
If the vaccine does work, scientists will face another challenge: determining who needs it. While detectable genetic flaws cause some Alzheimer's cases, most patients have no such telltale markers.
There is no known cure for Alzheimer's, which is believed to affect more than 4 million Americans. The patients, most of them older than 60, progress from forgetfulness to dementia and usually die within five to 10 years after diagnosis.
Neurobiologist Peter St. George-Hysop at the University of Toronto said the study raises the prospect of using immunization to treat or prevent other diseases associated with protein deposits.
Creutzfeldt-Jakob disease, the fatal human brain disorder thought to be related to mad cow disease, would be a candidate. So would Parkinson's disease and myeloma, a form of cancer that overproduces protein in bone marrow.