The UN Millennium Development Goals highlight the infant mortality rate (IMR) as a measure of progress in improving neonatal health and more broadly as an indicator of basic health care. However, prior research has shown that IMRs (and in particular perinatal mortality) can be underestimated dramatically, depending on a particular country's live birth criterion, vital registration system, and reporting practices. This study assesses infant mortality undercounting for a global dataset using an approach popularized in productivity economics. Using a one-sided error, frontier estimation technique, we recalculate rates and concurrently derive a measure of likely undercount for each country. © 2009 Elsevier Ltd. All rights reserved.