The Brazelton Neonatal Behavioral Assessment Scale was applied in 120 healthy full-term newborn infants. Interobserver reliability on day 5 (n = 20), test-retest reliability on day 4-5 (20), and correlations between examinations on day 1 and day 5 (80) were studied with regard to individual and grouped items (subscales) in the test. The overall interobserver reliability generally seemed to be satisfactory, but low estimates of test-retest reliability were obtained for many items. Items assessing orientation capacity and regulation of state showed better test-retest stability when the predominate states were identical on both occasions. Some subscales were significantly related to parity, sex, birthweight and age in hours at examination.