Decoding someone’s facial expressions can provide insights into their emotional experience. Recently, Automatic Facial Coding software was developed to provide continuous measurement of emotional facial expressions. Previous studies documented the sensitivity of a state-of-the-art system (Facereader [FR], Noldus Information Technology); it has detected facial responses in people who passively viewed emotional scenes. In the present experiment, we set out to generalize these results to affective responses under socially relevant intentions. Thus, we presented participants with pictures of facial expressions and instructed them to control their emotional expressions. Healthy participants (N = 64) viewed pictures with happy, neutral, or angry expressions. In a between subjects design, they were either instructed to actively mimic the expression, or to look at the picture passively, or to actively inhibit their own facial reaction. A video stream (FR) and EMG (zygomaticus and corrugator) were registered continuously. In the mimicking condition, both FR and EMG differentiated well between all emotions. In the passive viewing and in the inhibition condition FR did not detect changes in facial expressions whereas EMG was highly sensitive, even when participants intended to conceal their emotional responses. These data extend previous findings that automatic facial coding is a promising new tool for the detection of intense emotional expressions, but is not yet sensitive enough to detect more subtle reactions.