Diazepam has been broadly accepted as an anxiolytic drug and is often used as a positive control in behavioral experiments with mice. However, as opposed to this general assumption, the effect of diazepam on mouse behavior can be considered rather controversial from an evidence point of view. Here we revisit this issue by studying the effect of diazepam on a benchmark task in the preclinical anxiety literature: the elevated plus maze. We evaluated the minute-by-minute time-course of the diazepam effect along the 10 min of the task at three different doses (0.5, 1 and 2 mg/kg i.p. 30 min before the task) in female and male C57BL/6J mice. Furthermore, we contrasted the effects of diazepam with those of a selective serotoninergic reuptake inhibitor (paroxetine, 10 mg/kg i.p. 1 h before the task). Diazepam had no anxiolytic effect at any of the tested doses, and, at the highest dose, it impaired locomotor activity, likely due to sedation. Noteworthy, our results held true when examining male and female mice separately, when only examining the first 5 min of the task, and when animals were subjected to one hour of restrain-induced stress prior to diazepam treatment. In contrast, paroxetine significantly reduced anxiety-like behavior without inducing sedative effects. Our results therefore suggest that preclinical studies for screening new anxiolytic drugs should be cautious with diazepam use as a potential positive control.