Objective The specific objectives of this study were to examine variation in the care of self-harm patients in hospital settings and to identify the factors that predict recommended next care following self-harm. Methods Data on consecutive presentations to Irish emergency departments (EDs) involving self-harm from the National Self-Harm Registry Ireland from 2004 to 2012 were utilised. Univariate and multivariate regression analyses were performed to assess the associations between patients’ clinical and demographic characteristics, and recommended next care received. Results Across the study period a total 101,904 self-harm presentations were made to hospital EDs, involving 63,457 individuals. Over the course of the study there was a declining number of presentations resulting in patient admission following attendance with self-harm. Recommended next care varied according to hospital location, with general admission rates ranging from 11% to 61% across administrative health regions. Multinomial logistic regression identified that the factor which most strongly affected next care was the presenting hospital. Being male, older age, method, repeat self-harm, time of attendance and residence of the patient were all identified as influencing care received. Psychiatric admission was most common when highly lethal methods of self-harm were used (OR = 4.00, 95% CI, 3.63–4.41). A relatively large proportion of patients left the ED without being seen (15%) and the risk of doing so was highest for self-harm repeaters (1.64, 1.55–1.74 for those with 5+ presentations). Conclusions The extensive hospital variation in recommended next care indicates that management of self-harm patients may be determined more by where they present than by the needs of the patient. The study outcomes underline the need to standardise the clinical management of self-harm patients in general hospital settings.