The study investigated alexithymia in relation to facial imitation, empathy, attachment (models-of-self and -others), and reported emotional contagion and the degree of correspondence between facial responses (somatic affect) and reported emotional contagion (mentalization of affects). Pictures of angry, happy, and sad faces were presented to 102 participants at exposure times from subliminal (17–23 ms) to clearly supraliminal (2,500 ms). Electromyographic (EMG) activity from the corrugator and zygomaticus muscles was measured, and emotional contagion was reported after exposures. Self-report questionnaires measured alexithymia (TAS-20), attachment (RSQ), and empathy (IRI). Alexithymia correlated negatively with models-of-self (p < .01) and models-of-others (p < .0001), with a marginally significant correlation with empathy (p = .06). An interaction (p < .05) was found between alexithymia and facial expression with corrugator response as dependent measure, indicating corrugator imitation only for the low-alexithymia participants. The low-alexithymia group’s facial responses corresponded with their verbal reports, when exposed to both negative and positive faces at the longest exposure. The high-alexithymia participants showed correspondences when exposed to happy facial expressions at shorter exposures.