Laboratory activities require scientific processes. The crucial scientific process in laboratory activities is aligning with 21st Century skills. Argumentation skills in laboratory activities are needed to make laboratory activities run efficiently, effectively, and without misinformation. Sophisticated Thinking Blended Laboratory (STB-LAB) in laboratory activities is considered capable of improving argumentation skills because of its syntax, disposition, and argumentation stage, allowing students to train their argumentation skills. This study intends to determine whether the STB-LAB learning model can improve students' argumentation skills in virtual and real laboratory activities. The method uses quantitative research with a one-group pretest-posttest design in which the N-Gain will be generated from the pretest and posttest, and find the difference in pretest and posttest, t-test using independent sample t-test, and effect size using Cohen's D. The results show that N-Gain obtained in virtual laboratory activities, only two aspects were categorized as quite effective, namely warrant, and backing aspects. In addition, N-Gain obtained in real laboratory activities only has three aspects categorized as quite effective: evidence, warrant, and rebuttal. The hypothesis obtained in both laboratory activities is 0.000 at Sig. (2-tailed), and the Effect Size obtained in both laboratory activities is 0.91 and 0.79. Thus, STB-LAB can improve argumentation skills using two media, with the virtual laboratory getting the highest results in its improvement. This shows that STB-LAB is appropriate for improving argumentation skills at the undergraduate level.