The present paper proposes the category of 'indiscriminate homicide'. In a broad sense, it is a form of violence (1) towards stranger/s and (2) without identifiable rational motivation or unknown circumstance. Apart from the two elements, in a narrow sense, it should include only homicides committed (1) in a public place and (2) during daytime. Based on the above tentative definition, this paper estimated the prevalence of indiscriminate homicide in Taiwan is about 0.11 per 100,000 inhabitants in 2015. It is rather difficult to find the similar estimation in other countries. However, we could compare the prevalence of stranger homicide as a proxy measure of indiscriminate homicide between countries. It is estimated that 0.25 per 100,000 for Taiwan (including attempt cases), 0.39 per 100,000 for the USA and 0.22 for the UK comparatively speaking for their prevalence of stranger homicide. By comparing four types of homicide, namely Type A indiscriminate homicide, Type B stranger homicide with specific gain, Type C domestic and acquaintance homicide, and Type D mixture of stranger, family and acquaintance homicide, the study finds that there are significant differences among these four groups. The indiscriminate homicide cases are more likely to be committed during daytime, about zero likelihood of involving a female perpetrator, average age 40, mostly employed in labor or service jobs, low educational attainment, primary method of killing by knife, 5% suffering from mental illness or disability, and half of them involving advance planning. The paper then discusses the prevention challenges that face development of an appropriate social security network against such a rare and peculiar violent event. To profile, assess, and intervene on indiscriminate homicide will require prioritizing a small number of important parameters. We must recognize that even when we embed our social security network, the inherent limits of any predictive knowledge means that some further homicides will not be prevented.