This research aims to determine: 1) the significant difference between groups of students who were taught using the Discovery Learning, Problem-based Learning model using animated media and the group of students who were taught conventionally; (2) magnitude of the effectiveness of Discovery Learning and Problem-based Learning models to improve science learning outcomes in fifth grade of SD Negeri Gugus II Kuta Badung. This research used quasiexperimental research with the Non-equilevalent Control Group Design. The data were collected through test techniques. The research population was seven classes of fifth grade elementary school students in SD Negeri Gugus II Kuta Badung from five elementary schools. The research sample was determined using a random sampling technique consisting of three classes. Based on the results of the One-Way Analysis of Variance obtained Fcount = 17,855, Sig 0.000 <0.05. This means that there is a significant difference in science learning outcomes between groups of students who are taught using the Discovery Learning, Problem-based Learning model using animated media, and the group of students who are taught conventionally. Followed by calculating the effectiveness of the learning model to improve science learning outcomes using Partial Eta Squared, the results obtained Partial Eta Squared (η 2 )= 0,247. So it can be concluded that the application of the Discovery Learning, Problem-based Learning models using animated media is effective in improving science learning outcomes with an effectiveness of 0.247 in the medium category.