A systematic review of the literature about patients' preferences for involvement in cancer treatment decision making was conducted. Establishing preferences is important if the aim is to make health care more sensitive to the needs and expectations of each individual patient. Thirty-one papers were included in the review. Generalising from this literature is problematic because of limitations related to sample size, sample composition and methods used to assess preferences. Whilst we take cognizance of these limitations, research suggests that preferences vary considerably and that whilst most patients prefer a collaborative role, a significant minority prefer a passive or active role. Evidence about the association of factors such as age, gender, level of education, marital status, socio-economic status and health status with preferences is inconclusive. Only a handful of studies investigated the degree of congruence between patients' role preferences and the actual role that they perceived they had played, which highlight that some patients experience a dissonance between the two. Similarly, few studies investigated the impact of this dissonance on patient anxiety or satisfaction with the treatment decision. We advocate more rigorous investigations before recommendations for health care professionals can be processed with confidence.