Few studies have examined how the dynamic qualities of typical infant-caregiver interactions change after a perturbation or how they relate to the recovery process following perturbation. Further, these studies often present primarily static measures of dyadic interaction. This study examined dyadic affective flexibility, a dynamic measure of variability in the infant-caregiver affective interaction that provides a dyadic communication gestalt through synthesis of dyadic measures such as range, fluctuation, dispersion, stickiness and uncertainty, measures which have not yet been studied in the context of infant-caregiver interactions. The pre- and post perturbation face-to-face play interactions during the Face-to-Face Still-Face paradigm (FFSF), a well established perturbation of the face-to-face interactions that typify social engagement during infancy, afford a unique opportunity to study baseline and post perturbation dynamics of the infant-caregiver interaction. The first aim of this study was to characterize dyadic affective flexibility of 124 three month old infants and their mothers during the FFSF, before and after the still-face perturbation, by assessing the nature of change in dyadic flexibility using the state space grid technique. A second aim of this study was to study the relationship of dyadic flexibility during the initial play episode with infant negativity during the still-face episode. A third aim was to examine the relationship of dyadic flexibility with the quality of dyadic resolution during the reunion play episode. The effects of infant sex and maternal level of depressive symptoms were also explored. The main finding of the study was that dyadic flexibility in infant-mother dyads increases following the still-face perturbation (F = 13.32, p < .001, η² = .12). Further, there was a episode by infant sex by caregiver depressive symptoms interaction (F = 5.352, p = .022, η² = .043) such that the dyads with male infants and mothers with clinically significant depressive symptoms were most prone to reorganization of their interaction following the still-face perturbation, as were female infants of mothers with low levels of depressive symptoms. Baseline dyadic interactional measures were found to predict infants' negative affective response to perturbation. Developmental research and clinical implications are discussed.