The present study reconstructed a Sentence Completion test and scoring manual (Estrin, 1994) originally designed to measure K ohut and W olfs (1978) mirror-hungry, ideal-hungry, and twinship-hungry subclinical narcissistic personalities in a student sample. In order to extend the test’s range, 140 student and 20 psychotherapy client participants were recruited. Operational definitions were revised and subscales constructed according to theory-driven rules, resulting in a 24-item test with 3 subscales representing the behavioral expression o f the narcissistic need (ml, H, t1), and 3 representing narcissistic vulnerability (m2, i2, t2). Inter-rater reliability was satisfactory for all subscales except H. Student participants demonstrated greater idealhunger and grandiose narcissism. Females scored higher than males on narcissistic vulnerability. An analysis o f scoring patterns determined that, as expected, mirror-hunger was most prevalent, and twinship-hunger the least. A principal components analysis resulted in 4 oblique factors, suggesting both vulnerable and grandiose mirror-hungry narcissism, and the attraction and disappointment dynamics underlying twinship-hunger. Aggregated content validity findings were consistent with theoretical models for mirror- and twinship-hunger, but not ideal-hunger. All subscales except H demonstrated convergence with self-rating measures o f the personality constructs, but predicted associations between subscales and corresponding other-rating scales co m p leted by therapists o f clinical participants w ere n o t supported. Convergent validity was not demonstrated for the H subscale, whereas i2 was related to defensiveness. Twinship-hungry scales were related to dependency and reassurance-seeking (for t1), and risk-avoidance (for t2). The combination o f the m l and m2 subscales conformed to K ohut & W olfs (1978) description o f the mirror-hungry type as being both attention- and recognition-seeking, and defensive. As predicted, the m l (attention-seeking) subscale was relat d to overt narcissism, whereas the m2 (vulnerability) subscale was associated with both overt and covert characteristics. Product term multiple regression analyses were inconclusive in attempting to demonstrate: a) that the low m l / high m2 scoring profile was related to covert narcissism, and b) that narcissistic vulnerability (as manifested by the high m l / high m2 scoring profile) was associated with overt narcissism. However, cumulative evidence suggested a relationship between overt narcissism and narcissistic vulnerability. Overall findings supported the validity o f the mirror-hungry scale as a measure o f exhibitionistic-vulnerable subclinical narcissism.