A cancer clusters may be defined as an increased incidence of cancer which is restricted to a particular location and time period and is reported to health authorities. During the past decade, at least tens and probably hundreds of cancer clusters have been reported in Australia. Where warranted by circumstances such as a single tumour type afflicting individuals of an unexpectedly young age, investigation of clusters usually involves consideration of two matters. The first is whether the cases reported (typically 15 or less) represent an increased incidence of disease by reference to relevant registry data. The second is whether a plausible cause(s) is apparent with reference to potentially carcinogenic exposure(s) consequent upon relevant environmental circumstances. Some authorities favour such an environmental approach as the primary consideration. However, the approach is limited in its application because, for most tumour types, there are relatively few proven causal agents. Review of cluster investigations has consistently demonstrated that chance, rather than an insidious exposure to a known, or even an unknown carcinogen, is almost invariably specified as accounting for the cancer cases in question. However, the anxiety which prompts reporting of clusters by affected communities is rarely alleviated by references to chance. With the exception of lung cancer attributable to smoking, understanding by the wider community of cancer causation is at least incomplete and often wrong. The means of establishing agents that cause cancer, as well as the identity of such agents, are subject to erroneous beliefs. The situation is attributable to, and exacerbated by media reporting both in relation to clusters and also concerning carcinogenic risk consequent upon consumer products in particular. Indeed, there is evidence that concern about ‘cancer scares’ detracts from individual commitment to proven cancer prevention measures such smoking cessation. To that extent, improved community education about clusters may benefit many more people than the relative handful identified by a cluster report.