The study compared the responses of fourth- and ninth-grade children between equivalent inference patterns in class logic and conditional logic. A testing instrument of 72 class-logic items and 72 conditional-logic items was constructed for the study. Eighty-one fourth-grade students and 67 ninth-grade students participated. Results showed that for the fourth-grade students, differences between mean scores on the class logic and on the conditional logic were not statistically significant. For the ninth-grade students, differences between mean scores on the class-logic inference schemes with premises "no A are B" and the conditional-logic inference schemes with premises "if p(x), then not q(x)" were statistically significant. Differences between inference schemes with a premise"all A are B" and with "no A are B," differences between inference schemes with a premise "if p(x), then q(x)" and with "if p(x), then not q(x)," and differences between mean scores for items testing reasoning from the converse and testing logical equivalency of a statement and its converse were all statistically significant for both fourth- and ninth-grade students. (Author/DT)