(As always, I post fascinating stuff I learn at work, and I try to put it in simple terms as much as I can. Please feel free to add/clarify in the comments as you see fit.)
The September issue of the Cancer Journal is dedicated to cancer vaccines and how they may hold the key for cancer treatment and prevention. This is not to be confused with vaccines against cancer-causing viruses, like HPV. In that case the vaccine elicits antibody responses against the virus. In the context of cancer, though, a vaccine would use the immune system's own weapons in order to destroy tumor cells. An example is the vaccine to treat advanced prostate cancer that was approved by the FDA in April 2010, after a Phase III trial showed that patients who received the treatment survived longer than the controls.
The main question in order to create a vaccine that targets cancer cells is: how do we tell the immune system which are the cells to destroy? There is a particular class of cancer cells that offers a potential candidate: cancer stem cells.
Stem cells are a class of very special cells because they are undifferentiated, which means they have the potential to generate any kind of tissue (heart, lung, skin, etc.) Seems almost a paradox, doesn't it? Stem cells can remain undifferentiated and at the same time differentiate into specialized tissue cells. This is possible through an asymmetric cell division: every time a stem cell divides, it generates two cells, an undifferentiated stem cell, and a differentiated one. This way, the differentiated cells produce the specialized tissue, while the stem cell population remains intact.
In a healthy individual, cells with this capacity are found in the bone marrow and in umbilical cords. Unfortunately, they have also been found in solid tumors, such as breast cancer, prostate cancer, and melanoma. You can immediately see the problem: if a cancer cell remains undifferentiated, it means it can preserve its population while generating new cancers in other parts of the body -- the process known as metastasis.
Therefore, one way to produce a cancer vaccine is to have it elicit immune responses against cancer stem cells . How? The idea is to use proteins, or even bits of proteins (peptides) that are over-expressed on tumor cells. Vaccines that use peptides as antigens are called anticancer peptide vaccines , and right as I was reading about them, one of the authors of this paper  wrote this wonderful article on Scientific American, which describes in great detail the history and ideas behind a cancer vaccine. Quoted from the S.A. article:
"Basically, there are three elements to making a cancer vaccine. The first is to decide precisely what molecular feature, or antigen, in a malignant tumor the immune system should recognize as foreign and target for killing. The second is to decide how to deliver a triggering agent (or vaccine) to the immune system that ramps it up to attack cancer cells. And the third is to decide which cancer patients to treat and when during the course of their disease to administer the vaccine."
Mutated cancer cells arise normally (in small quantities) in the body and a healthy immune system is normally capable of recognizing them and destroying them. A vaccine would make this kind of response stronger and robust enough to wipe out all malignant cells. Unfortunately, as cancer progresses, the immune system gets severely damaged. Therefore, the key for this strategy would be to either act fast enough (when the tumor is still small), or combine it with other strategies like chemotherapy.
In the September issue of the Cancer Journal, Dhodapkar et al.  review what the future holds in cancer vaccine research, whereas Larocca et al.  discuss viral vectors, in other words, how viruses could be engineered to deliver a cancer vaccine.
 Dhodapkar MV, & Dhodapkar KM (2011). Vaccines targeting cancer stem cells: are they within reach? Cancer journal (Sudbury, Mass.), 17 (5), 397-402 PMID: 21952290
 Perez SA, von Hofe E, Kallinteris NL, Gritzapis AD, Peoples GE, Papamichail M, & Baxevanis CN (2010). A new era in anticancer peptide vaccines. Cancer, 116 (9), 2071-80 PMID: 20187092
 Larocca C, & Schlom J (2011). Viral vector-based therapeutic cancer vaccines. Cancer journal (Sudbury, Mass.), 17 (5), 359-71 PMID: 21952287