The need for assessment of social skills in clinical and neurotypical populations has led to the widespread, and still increasing use of the 'Reading the Mind in the Eyes Test' (RMET) developed more than two decades ago by Simon Baron-Cohen and colleagues for evaluation of social cognition in autism. By analyzing most recent clinical and brain imaging data, we illuminate a set of factors decisive for using the RMET. Converging evidence indicates: (i) In neurotypical individuals, RMET scores are tightly correlated with other social skills (empathy, emotional intelligence, and body language reading); (ii) The RMET assesses recognition of facial affect, but also heavily relies on receptive language skills, semantic knowledge, and memory; (iii) RMET performance is underwritten by the large-scale ensembles of neural networks inside and well-outside the social brain; (iv) The RMET is limited in its capacity to differentiate between neuropsychiatric conditions as well as between stages and severity of a single disorder, though it reliably distinguishes individuals with altered social cognition or elevated pathological traits from neurotypical persons; (v) Merely gender (as a social construct) rather than neurobiological sex influences performance on the RMET; (vi) RMET scores do not substantially decline in healthy aging, and they are higher with higher education level, cognitive abilities, literacy, and mental well-being; (vii) Accuracy on the RMET, and engagement of the social brain, are greater when emotions are expressed and recognized by individuals with similar cultural/ethnic background. Further research is required to better inform usage of the RMET as a tool for swift and reliable examination of social cognition. In light of comparable visual input from the RMET images and faces covered by masks due to COVID-19 regulations, the analysis is of value for keeping efficient social interaction during the current pandemic, in particular, in professional settings related to social communication.