We examined the hypothesis that best friends and members from a broader peer group would not differ in the amount of influence they have on young adults' alcohol consumption and that what counts would be the mere presence of drinking peers in a given context--irrespective of the type of relationship such peers would have with the target young adult. Data were used from a naturalistic observation study that was conducted in a "bar-lab" among 221 young adults aged 18-25 years. Both hierarchical and multilevel regression analyses showed that group effects (i.e., average group levels of alcohol consumption) explained young adults' drinking. When taking into account these group effects, best friends' alcohol consumption in the bar-lab was a nonsignificant predictor of young adults' alcohol use--although best friends' questionnaire-assessed drinking "did" predict alcohol consumption in the bar-lab.