Two Japanese causal connectives, "kara" and "node," are often assumed by linguists to share many distributional similarities. This paper argues that they are in fact based on semantically or logically different assumptions. The paper reviews some past treatments of the connectives and suggests an alternative analysis in terms of the differences with respect to their logical structures involving assertion and presupposition. It is stated that "kara" occurs in a bisentential structure in which either: (1) S1 and S2 are both asserted where S2 contains some epistemic qualifier, or (2) only S1 is asserted and there is a related pseudocleft. However, "node" appears in a monosentential structure where S1 is presupposed such that if "p," then "q," and S2 asserted, with the speaker also presupposing that the hearer shares his presuppostion. (CHK)