This experimental research aims to find out whether or not: 1) the probing-prompting learning model affects learning achievement, 2) the self-efficafy affects learning achievement, and 3) the model affects the interaction between the learning model and student’s self-efficacy on learning outcomes. The population in this study were all third grade students of Christian Elementary School Eben Haezer Salatiga as many as 72 students. The sampling technique used Cluster Random Sampling and obtained by grade IIIA students as experimental class used probing-prompting learning model and class IIIC as control class using conventional learning model, with each consists of 24 students. Both models are used in science learning on natural resource materials. The instuments used were the test of science learning achievement and questionnaire about student self-efficacy. Data analysis included descriptive and inferential analysis consisting of normality test, homogenity test with Levene's, early equilibrium test with independent sample t test, and hypothesis test with Anava. The whole test was performed with SPSS calculation tool at 0.05 significance level. The results of the hypothesis testing concluded: 1) there was no effect of probing prompting learning model on students' learning achievements; 2) there was an effect of self-efficacy to the learning achievements; the result of student learning with high self-efficacy category is better (higher) than the moderate and low self-efficacy, and student’s learning result with moderate self-efficacy category was better than the ones with low self-efficacy; 3) There was no effect on interaction between the learning model and self-efficacy on student science learning outcomes.