Based on a Japanese longitudinal study, three assumptions underlying use of the ‘Strange Situation’ procedure are examined with respect to their universal validity. Sixty 12-month-old infants were assessed by means of the original procedure and their development followed until their 42nd month. Contrary to the key assumptions, (a) the stress aroused by the procedure went well beyond the moderate level for the Japanese infants, and some were pushed from type B to type C behavior by the procedure; (b) infants did not clearly exhibit avoidant behaviors toward the mother at the reunion with her, resulting in no type A babies; and (c) the insecurely attached type C infants overcome their ‘disadvantage’ by the 32nd month at the latest. These ‘dissonant’ findings are interpreted in terms of Japanese customs of child-rearing and interpersonal interactions.