The construct validity of four empathy instruments were evaluated through examining group differences and correlations between tests. Subjects were 18 registered nurses (RNs) and 32 nursing assistants (NAs). Instruments used were the Empathy Test (Layton, 1979), the Carkhuff (1969a, 1969b) Empathic Understanding in Interpersonal Processes Scale, the empathy subtest of the Barrett-Lennard (1962) Relationship Inventory, and LaMonica's (1981) Empathy Construct Rating Scale (ECRS). The first hypothesis, that RNs would score significantly higher than NAs on all empathy measures, was confirmed (p less than .0003 to p less than .027). The second hypothesis, that the various empathy measures would be positively related, was partially confirmed. The Carkhuff scale was significantly correlated with the ECRS (r = .25, p less than .05). The ECRS and the Barrett-Lennard also were correlated (r = .78, p less than .001). Results are interpreted using Barrett-Lennard's (1981) theory of empathy.