Objective: The purpose of this study was to compare the effectiveness of the four-component, Gagne instructional design models with each other and conventional method in experimental science lesson learning (complex subjects).Method: To achieve this purpose, the experimental method with pre-test, post-test, and control groups were used. The entire boys students of Karaj city were the population of this research. The sample was selected through cluster sampling that includes 60-mentioned population that divided into three groups (experimental1, exprimental2, and control group). Intended subject matter presented for Experimental 1 through four component instructional design model, experimental 2 through Gagne instructional design model and for control group through conventional method. For all three groups, pretest, posttest test was executed. For descriptive data analysis, mean and standard deviation scores have been used and for the inferential statistic, covariance analysis was used to examine the hypothesis.Results: Data were analyzed using descriptive statistics and covariance test. The findings showed that the mean of the control group and the tests in the pre-test was not significantly different, but in the post-test, the mean of the experimental groups was higher than the control group and by controlling the pre-test score, the observed differences between the scores of the experimental and control groups were significant.Conclusion: The results showed that the difference between the traditional education groups, four components and Gagne instructional design models with 95% confidence was significant. In this way, the Gagne instructional design model from the conventional method, and the four components instructional design model from Gagne instructional design model are more effective in science learning (complex subject).