The present study aims at examining the validity of Akatsuka and Clancy (1993) and Solvang (2004a, b), who made contradictory claims about the development of conditional expressions in Japanese. Focusing on forms and meanings, the use and development of three conditional expressions (-tara, -to, -ba) found in four infants' longitudinal data are analyzed in two studies. The results showed the following common tendencies among the children; (1) conditional expressions initially appear as a specific form with a specific function, indicating a single form-function mapping. (2) The use of conditional expressions spreads from the demotic modality (D conditional) to other forms of conditionals, even though the data characteristics are different between the two studies. These results support Akatsuka and Clancy's claim (1993) but not Solvang (2004b), in that the Japanese infant's conditional expression starts from a specific meaning function and the linguistic form like D conditional acquisition, and usually extends to to more diverse conditional forms and functions.