Social distancing norms have been promoted after the COVID-19 pandemic. In this work, we tested interpersonal space (IPS) in 107 subjects through a reaching-comfort distance estimation task. In the main experiment, subjects had to estimate the comfort and reach space between an avatar wearing or not wearing a face mask. We found that IPS was greater between avatars not wearing a mask with respect to stimuli with the mask on, while reaching space was not modulated. IPS increment in the NoMask condition with respect to the Mask condition correlated with anxiety traits, as shown with the State-Trait Anxiety Inventory, rather than with transient aspects related to the pandemic situation. In the control experiment, the avatars with a mask were removed to further explore the conditioning effect provided by the presence of the facial protection in the main experiment. We found a significant difference comparing this condition with the same condition of the main experiment, namely, the distances kept between avatars not wearing a mask in the main experiment were greater than those between the same stimuli in the control experiment. This showed a contextual adaptation of IPS when elements related to the actual pandemic situation were relevant. Additionally, no significant differences were found between the control experiment and the Mask condition of the main experiment, suggesting that participants had internalized social distancing norms and wearing a mask has become the new normal. Our results highlight the tendency of people in underestimating the risk of contagion when in the presence of someone wearing a mask.