An adjuvant is defined as a product that increases or modulates the immune response against an antigen (Ag). Based on this general definition many authors have postulated that the ideal adjuvant should increase the potency of the immune response, while being non-toxic and safe. Although dozens of different adjuvants have been shown to be effective in preclinical and clinical studies, only aluminium-based salts (Alum) and squalene-oil-water emulsion (MF59) have been approved for human use. However, for the development of therapeutic vaccines to treat cancer patients, the prerequisites for an ideal cancer adjuvant differ from conventional adjuvants for many reasons. First, the patients that will receive the vaccines are immuno-compromised because of, for example, impaired mechanisms of antigen presentation, non-responsiveness of activated T cells and enhanced inhibition of self-reactivity by regulatory T cells. Second, the tumour Ag are usually self-derived and are, therefore, poorly immunogenic. Third, tumours develop escape mechanisms to avoid the immune system, such as tumour editing, low or non-expression of MHC class I molecules or secretion of suppressive cytokines. Thus, adjuvants for cancer vaccines need to be more potent than for prophylactic vaccines and consequently may be more toxic and may even induce autoimmune reactions. In summary, the ideal cancer adjuvant should rescue and increase the immune response against tumours in immuno-compromised patients, with acceptable profiles of toxicity and safety. The present review discusses the role of cancer adjuvants at the different phases of the generation of antitumour immunity following vaccination.