BACKGROUND: To study specific effects of four maternal risk factors: age, parity, educational level, smoking, for specific causes of stillbirth and neonatal death according to a previously described hierarchic classification. METHODS: The study is based on 9,785 stillbirths or neonatal deaths among infants born in Sweden, 1983-1995 (n=1,412,754) and identified with various Swedish health registers. Statistical analysis is performed using Mantel-Haenszel analysis. RESULTS: Some risk factors, known from the literature, were confirmed and could be quantified. In addition, high parity was shown to increase the risk for death associated with multiple births (OR=2.49, 95% CI 2.07-3.01) and low educational level seems to be protective for such death (OR=0.75, 95% CI 0.60-0.93). If the infant is SGA, the risk for death is higher at high than at low parity (1.70, 95% CI 1.19-2.43, and 1.0, 95% CI 1.06-1.15, respectively). Maternal smoking seems to aggravate the placental abruption because the death risk in the presence of abruption increases when the mother smoked (OR = 1.74, 95% CI 45-2.08). CONCLUSIONS: The study shows that the groups of the classification system used (NICE) differ in their association with known risk factors for stillbirth and neonatal deaths and an analysis based on specific causes of death can therefore unravel risk factors hidden when total mortality is used. The computerized method of classification and the cause-of-death classification developed by us is clearly useful for such analyses which requires large materials.