This study seeks to enrich our understanding of modeling‐based learning (MbL) in kindergarten science education, investigating the influences of different modeling tools on kindergarten child‐constructed models and their modeling reasoning. Therefore, this multi‐case study aimed at providing in‐depth descriptions of how MbL was enacted by 66 kindergarteners while combining the use of three modeling tools: paper‐and‐pencil, three‐dimensional structures, and dramatic play. We studied three different classes of children engaged in MbL who studied and modeled three different physical phenomena (wildflowers' parts and their functions, dissolving substances in water, shadow formation). We varied the modeling tools to investigate the ways these tools contributed to children's MbL. Data sources included the child‐constructed models, their modeling discourse, and mechanistic reasoning. From the videotaped lesson transcripts, we developed detailed accounts of the three MbL cases, and we analysed the discourse using modeling frame analysis and mechanistic reasoning analysis, along with the model component artifact analysis of the child‐constructed models. Findings suggested that the use of different modeling tools impacts child‐constructed models and their corresponding modeling‐based mechanistic reasoning. The findings also suggest that children's representational proficiencies seem to be related to the use of a variety of modeling tools, which enabled children to talk about the possible different ways of developing models and the different affordances of the various modeling tools' representational power. We use this evidence to argue that different modeling tools may afford different modeling possibilities that kindergarten children may draw upon possibly combining with modeling resources they have (e.g., role‐playing, storytelling, drawing).