A review of the literature on suicide with particular emphasis on its research aspects has been completed. Demographic studies have given some information about the circumstances surrounding suicide, but have been criticized for their incomplete data, the validity of their statistics, and their failure to explain individual suicide. Psychodynamic studies have relied heavily on the dynamics of depression to explain suicide and generally have failed to explain suicidal behaviour in other dynamic constellations. Clinical studies searching for reliable indicators of suicide have encountered many methodological problems making conclusions uncertain, thus limiting practical application of the findings. An important neglected area of study is felt to be the phenomenology and developmental psychology of suicidal ideation. Some of the sparse literature in this area is reviewed and its significance in terms of object relations theory is discussed. The hypothesis that the suicidal impulse is a variant of normal psychological development with important determinants in early childhood is proposed. This proposal is discussed in terms of its implications for future research.