One of the most influential factors impacting the development of a young person's security in relation to self is the reciprocal interpersonal relationship between child and parent. In early childhood in particular, the interactions between infant and parent serve as an organizing framework for how a youngster ultimately comes to integrate emotions and behavior in relation to her emerging sense of self, as well her growing ability to empathize with others. Such connections have crucial implications across the developmental trajectory. One area in particular that is receiving growing attention is the caregiver's capacity to reflect upon the mental experiences of self and other. This capacity has been demonstrated to influence the emotional quality and depth of relationships in terms of feelings of intimacy, emotional connectedness, and healthy individual autonomy, allowing for increased access to and meaning-making of a wider range of affective experiences (Fonagy et al., 2002). The primary focus of this exploratory study was to develop a measure of parental reflective functioning and parental self-reflection. It also examined the degree to which group differences existed in caregiver classifications based on this measure. Drawing on attachment theory, parent-child relationship research, and research on metacognitive monitoring and reflective functioning, Gilbert, Whelan, Marvin, & Stewart (2007) developed the Caregiver Reflective Functioning Scales (CRFS) designed to measure reflective functioning for specific use with the Circle of Security Intervention (COSI) project. Five cases (from a total n=8 1) of caregiver/child dyads who participated in the pre-intervention phase of the COSI were selected based on purposeful maximal sampling to ensure prototypes of caregiver classifications. Each case interview was coded using the CRFS. Results indicated the CRFS items most frequently observed when stratified by caregiver classification generally fit the conceptualization of the caregiving patterns that has been described in previous research. Findings also revealed the parental scale to yield a substantially higher number of coding observations than the self scale. Within these observations for both scales, the majority of responses were coded as "did not observe". A review of the results and limitations from the present study, as well as proposed directions for future research is provided.