Prior research has shown that the cognitive interview (CI) can be used to elicit a more complete and accurate report from eyewitnesses, but the bulk of the research has involved testing witnesses with high levels of education. In the present experiment, we examined whether the CI was effective with witnesses who were not college educated. Employees of the cleaning service of a University in Brazil witnessed a simulated armed abduction. They were then interviewed with a CI or standard interview (SI). Replicating earlier research, there was a significant increase in the number of correct details reported when witnesses were interviewed using CI with no cost to accuracy. This experiment provides the first test of the CI in a developing country and the implications for police interviewing practices in Latin America will be discussed. Copyright © 2006 John Wiley & Sons, Ltd.