Police investigators, judges, and jurors are often very skeptical of alibi witness testimony. To investigate when and why individuals lie for one another, we conducted two studies in which witnesses' support of a false alibi was observed. We varied the level of social pressure exerted on witnesses and the level of affinity between suspect–witness pairs. During a study session purportedly intended to investigate dyadic problem‐solving ability, a mock theft was staged. When questioned, participants were provided the opportunity to either corroborate or refute a confederate's false alibi that the latter was with them when the theft occurred. Participants were more likely to lie for the confederate when the latter explicitly asked participants to conceal his/her whereabouts during the time of the theft (Study 1). How much participants liked the suspect did not impact lying; however, participants lied for a confederate more often when the latter was a friend rather than a stranger (Study 2). Results show that alibi witnesses often lie and that investigators and jurors may not accurately estimate the likelihood that such witnesses will lie for one another. Witnesses who lied also reported doing so more often because they believed that the suspect was innocent rather than guilty. Copyright © 2016 John Wiley & Sons, Ltd.