This study explored questions concerning the specificity of facial expression of emotions in infants. A sample of 27 infants (nine each at ages 10, 11, and 12 months) were tested in six experimental and six control conditions. Two of the experimental conditions were designed to elicit fear, two happiness, and two surprise. Videotapes of the infants' responses were subjected to three different types of rating tasks with different sets of raters assigned to each. The three rating tasks were: assessment of instrumental behaviors indicative of emotional expression; forced-choice identification of infant emotion from facial expression alone; and judging the occurrence of component behaviors in different areas of the face. Results indicated that: (1) the judges were able to make all infant facial expression judgments reliably; (2) data from instrumental behaviors confirmed that all eliciting circumstances except the stranger approach were successful in producing the desired affective reactions in the infants; (3) forced-choice judgments for the control conditions were accurate with the exception of the visual cliff control episode; (4) forced-choice judgments were very accurate for happiness but not for the other experimental conditions; and (5) except for happiness, few of the facial expression components differentiated between experimental and control conditions. (JMB)