Mathematical models are necessary to understand the mechanisms of development, because the development of multicellular organisms is a spatially and temporally dynamic phenomenon. To develop a predictive mathematical model for such a dynamic phenomenon, we need to measure the target phenomenon, develop mathematical models based on the measurement, simulate the models’ dynamics, validate the models by comparing models’ and real dynamics, and repeat this process. To develop predictive models of development, we are developing technologies for the measurement, modeling, simulation and model validation, by integrating molecular cell biology, genomics, biophysics, computational science, informatics, mathematical science etc. We are taking two complementary approaches: 1) top-down approach, generating coarse-grained models and fine-graining them, 2) bottom-up approach, generating basic building-block models and integrating them. We are developing basic technologies and models in C. elegans, the simplest multicellular model organism, and applying them to mammalian model organisms.