This study explored the experimental effects of sex education program for elementary school students. The study adopted a nonequivalent pretest-posttest design, and the subjects were two 5th grade classes in Kaohsiung City. The members of experimental and control group are 28, respectively. The latter group didn't accept any topic about sexual teaching, but the former group was taught a total of 7 topics over the course of 14 classroom hours. The two groups were measured four times using the same Sexuality Attitude Scale, and the data were analyzed with SPSS 16.0 and Amos16.0. The results were as follows: (1) The program had both immediate and delayed effects on the sexual attitudes of the experimental group. (2) The experimental group demonstrated the existence of latent growth factors. (3) The experimental group showed a ceiling effect. (4) The latent growth curve model for the experimental group showed an upward trend followed by a downward one. Based on these conclusions, four recommendations are offered: (1) When teaching different sexual attitudes, teachers may refer to the sex education program designed by this study. (2) Teachers should continue to mention issues having to do with sexual attitudes after the experiment. (3) The cross-validity of the latent growth curve model needs to be retested.