Objectives: To identify factors underlying the reduction in neonatal mortality in the Czech Republic during the 1990s and to identify scope for further improvements. Design: Examination of trends in birth weight and birth weight specific neonatal mortality in the Czech Republic and comparison with figures from Sweden, which has one of the lowest neonatal mortality rates in Europe. Setting: The Czech Republic. Subjects: All singleton births occurring in the Czech Republic in 1989–1991 and 1994–1995, with a comparison group of all singleton births in Sweden in 1989–1991. Main outcome measures: Neonatal mortality rate. Results: Despite a slight worsening in the birth weight distribution, the neonatal mortality rate in the Czech Republic fell from 5.6 to 3.8 per thousand live births. This was due to an improvement in the survival of infants at all birth weights but especially amongst the lightest. Eighty percent of the overall improvement was due to greater survival among those under 2500 g. Comparison with Swedish birth weight specific rates indicates that, for the country as a whole, only small additional gains are likely as a result of improved survival at a given birth weight but, instead, a reduction in the proportion of low birth weight babies would have a much greater effect. Regional analysis indicates that the improvements have been much greater in Prague than in the rest of the country. Conclusions: An effective strategy to reduce neonatal mortality in the Czech Republic should have two elements. The first is to address the socio-economic determinants of low birth weight. The second is to reduce regional inequalities in the quality of neonatal care. This should, however, be supplemented by more detailed investigation to identify specific amenable factors.