This study looks at innovative learnings' effect on achieving students' High Order Thinking Skills (HOTS). HOTS is one part of the skills that need to be developed in the 21st Century. This research is a quantitative study with a meta-analysis approach to the mean with random-effects models. Generally, the steps involve formulating research problems meta-analysis to be performed, collecting studies, calculating the effect size, test of heterogeneity, data analysis (summary effect, forest plots, funnel plots, test the potential for publication bias), and concludes. The research sample of 42 research includes international journals and national, international, and nationwide proceedings and a thesis student. The analysis using random models (random-effect model) obtained a mean value of the effect size of aggregation or summary effect amounted to 77.37, standard error weighs the mean 12.36, the upper limit of 53.14, and the lower limit of 101.60. Analysis of the null hypothesis (Ho: actual effect [theta] = 0) leads to rejecting Ho because the Z value of the summary effect is 77.37 with the one-tailed p-value (0.00) smaller than the [alpha] value (0.05). Analysis of publication bias tests from Funnel Plot, Rank Correlation, Regression Method outputs, and "Trim and Fill" shows no potential publication bias regarding conclusions drawn. So the findings made based on the random effect model on the application of innovative learning affect HOTS students achievement (analyzing, evaluating, creating, critical and creative thinking skill aspects) are valid. The implication is that teachers or other practitioners can use innovative learning to improve students' HOTS.