Scratching has been widely used as an indicator of anxiety in many primate species. However, a handful of studies have shown no change in scratching under anxiety-provoking circumstances. In addition, the existing literature has investigated scratching only in relation to negative arousal (i.e., anxiety), even though anxiety and excitement (positive arousal) share important physiological and behavioral correlates, including increased heart rate, blood pressure, and locomotion. In the current study, we scored all instances of scratching in 11 outdoor-housed captive common marmosets (Callithrix jacchus) during three contexts that were intended to be negatively arousing and three contexts that were intended to be positively arousing during a baseline, manipulation, and post-induction period. Summed across the three negative arousal contexts, the results showed that subjects exhibited significantly lower scratching rates during the manipulation than during either the baseline or post-induction periods, and the pattern of means was the same for all three of those contexts. Under the three contexts of positive arousal, subjects exhibited different patterns of scratching rates during the manipulation periods (play = increases, foraging = decreases, food anticipation = no change). Data from the current study, and a close examination of data from studies showing no change in scratching under anxiety-provoking circumstances, suggest that the anxiety-scratching relationship may be more complex than has been reported previously. Our results raise a potential concern about the unchallenged use of scratching as a behavioral indicator of anxiety in captive non-human primates, with important implications for welfare and management of these species.