BACKGROUND: Primary intracerebral hemorrhage accounts for the relative minority of all strokes and yet is more fatal and disabling. Various prognostic models for mortality and functional outcome following primary intracerebral hemorrhage have been proposed, however there is little data which focuses on a multi-racial population profile characteristic of communities in South-East Asia. A reliable grading scale for this condition will allow for accurate risk stratification, treatment selection, resource allocation and possibly also aid in the definition of common enrollment criteria for clinical trials. METHODS: This study investigates an Asian population of primary intracerebral hemorrhage patients and defines using a variety of data mining techniques the clinical variables that significantly impact on early mortality. The models produced are then compared to ascertain which one optimally predicts this outcome. FINDINGS: Past history of stroke, known atrial fibrillation, use of warfarin, glucose level, presenting Glasgow Coma Scale (GCS) and pupil abnormality, post-resuscitation GCS and pupil abnormality, initial international normalized ratio (INR) and prothrombin time (PT) results, vomiting, seizure, total volume of clot, ventricular extension and hydrocephalus were significantly associated with early mortality. Logit with backward elimination showed that only age, presenting GCS, 1st INR result and total volume of clot were significantly associated with mortality in the final multivariate model. The use of the other data mining techniques yielded comparable results. CONCLUSIONS: The predictors for early mortality and poor outcome in primary intracerebral hemorrhage are similar in Asian and Western populations.