Preterm birth is the second largest direct cause of child deaths in children younger than 5 years. Yet, data regarding preterm birth (<37 completed weeks of gestation) are not routinely collected by UN agencies, and no systematic country estimates nor time trend analyses have been done. We report worldwide, regional, and national estimates of preterm birth rates for 184 countries in 2010 with time trends for selected countries, and provide a quantitative assessment of the uncertainty surrounding these estimates.

We assessed various data sources according to prespecified inclusion criteria. National Registries (563 datapoints, 51 countries), Reproductive Health Surveys (13 datapoints, eight countries), and studies identified through systematic searches and unpublished data (162 datapoints, 40 countries) were included. 55 countries submitted additional data during WHO's country consultation process. For 13 countries with adequate quality and quantity of data, we estimated preterm birth rates using country-level loess regression for 2010. For 171 countries, two regional multilevel statistical models were developed to estimate preterm birth rates for 2010. We estimated time trends from 1990 to 2010 for 65 countries with reliable time trend data and more than 10 000 livebirths per year. We calculated uncertainty ranges for all countries.

In 2010, an estimated 14·9 million babies (uncertainty range 12·3–18·1 million) were born preterm, 11·1% of all livebirths worldwide, ranging from about 5% in several European countries to 18% in some African countries. More than 60% of preterm babies were born in south Asia and sub-Saharan Africa, where 52% of the global livebirths occur. Preterm birth also affects rich countries, for example, USA has high rates and is one of the ten countries with the highest numbers of preterm births. Of the 65 countries with estimated time trends, only three (Croatia, Ecuador, and Estonia), had reduced preterm birth rates 1990–2010.

The burden of preterm birth is substantial and is increasing in those regions with reliable data. Improved recording of all pregnancy outcomes and standard application of preterm definitions is important. We recommend the addition of a data-quality indicator of the per cent of all live preterm births that are under 28 weeks' gestation. Distinguishing preterm births that are spontaneous from those that are provider-initiated is important to monitor trends associated with increased caesarean sections. Rapid scale up of basic interventions could accelerate progress towards Millennium Development Goal 4 for child survival and beyond.

Bill & Melinda Gates Foundation through grants to Child Health Epidemiology Reference Group (CHERG) and Save the Children's Saving Newborn Lives programme; March of Dimes; the Partnership for Maternal Newborn and Childe Health; and WHO, Department of Reproductive Health and Research.