This study aims to see how much influence the constructive play method has on the ability to recognize geometric shapes in early childhood. The method used is a quasi-experimental model of non-equivalent pretest posttest control group design. The data that has been collected, the researcher analyzes it using normality, homogeneity, and hypothesis tests. enough. The results of the calculation of the hypothesis test obtained a tcount of 4.09 > ttable of 2.093, which means that Ha is accepted, so there is a difference between those using the constructive play method and the storytelling method. Improved understanding of the geometry of the children in the control group obtained a value of 0.22 with low criteria, while the experimental group obtained a value of 0.66 with high criteria. The high understanding of children's geometry is caused by the influence of the constructive play method which obtains an effect size value of 1.79 > 1.00, which means it has a strong influence.