Maternal Mortality Rate in Indonesia is still high, around 230/100,000 live birth in 2005. Eclampsia is the second most cause of maternal death (about 13%). We have a very high prevalence of eclampsia in our center (Surabaya) about 1.08% of all delivery. One of our main problem about eclampsia case is multiple referral hospital system, because we do not have many tertiary center for this case. We assumed that high incidence of maternal death because of eclampsia is caused by this factor. To evaluate relationship between hospital referral system and maternal condition of eclamptic woman with maternal death event and develope scoring system to predict maternal death event on eclamptic woman. This is a case control study with retrospective observational analytic design, involve all eclamptic woman whom reffered and admitted to Soetomo hospital (Indonesia) from 1 January 2005 until 31 December 2009. Total eclamptic woman was 160 cases, with 12 maternal death on case group and 148 survive maternal on control group. Research variable includes maternal condition (ages, parity, eclampsia type, complication) and hospital referral pattern (ANC, Magnesium Sulfate, anti-hypertensive drugs, multiple hospital referral, health workers accompanion during patient referral). Primary outcome includes maternal death. There is significant association between multiple hospital referral (RO 6.1, P=0.083), maternal age >35 (RO 17.4, P=0.032), lung oedema (RO 19.1, P=0.065), renal failure (RO 98.8, P=0.002), intra cranial hemorrhage (RO 6319.7, P⩽0.0001), Eclampsia type (RO 0.009), on eclamptic women with maternal death. No significant association between parity (RO 0.3 P=0.310), HELLP syndrome (RO 2.2 P=0.478), Magnesium Sulfate given before patient reffered (RO 0.7, P=0.685), ANC quantity (RO 0.9 P=0.844), health workers accompanion during referral (RO 0.9, P=0.901), on eclamptic women with maternal death event. There are 6 significant variable includes Maternal complication (ICH, renal failure, lung oedema), Maternal ages >35 years, antihypertensive drugs, multiple hospital referral, multiparity, and eclampsia type that can be made a scoring system. We develop scoring system based on risk factor that can be used to predict maternal death event on eclamptic woman.