The nematode Caenorhabditis elegans has been extensively used as a model multicellular organism to study the influence of osmotic stress conditions and the toxicity of chemical compounds on developmental and motility-associated phenotypes. However, the several-day culture of nematodes needed for such studies has caused researchers to explore alternatives. In particular, C. elegans embryos, due to their shorter developmental time and immobile nature, could be exploited for this purpose, although usually their harvesting and handling is tedious. Here, we present a multiplexed, high-throughput and automated embryo phenotyping microfluidic approach to observe C. elegans embryogenesis after the application of different chemical compounds. After performing experiments with up to 800 embryos per chip and up to 12h of time-lapsed imaging per embryo, the individual phenotypic developmental data were collected and analyzed through machine learning and image processing approaches. Our proof-of-concept platform indicates developmental lag and the induction of mitochondrial stress in embryos exposed to high doses (200mM) of glucose and NaCl, while small doses of sucrose and glucose were shown to accelerate development. Overall, our new technique has potential for use in large-scale developmental biology studies and opens new avenues for very rapid high-throughput and high-content screening using C. elegans embryos. Microfluidics: High-throughput screening of embryosA new microfluidic approach enables the high-throughput and automated phenotyping of C. elegans embryos. Understanding the effect of osmotic stress-induced damage on cells is vital for understanding a number of biological processes. C. elegans embryos are considered ideal model systems for osmotic studies, but their phenotyping is traditionally a labor intensive and time-consuming process. Now, a team from EPFL in Switzerland led by Martin Gijs demonstrate a microfluidic platform for high-throughput phenotyping of C. elegans. Embryos are exposed to various osmotic compounds, followed by an automated phenotyping script to assess phenotypes via machine learning and image processing. Their proof of concept experiment demonstrates that their technique can be used for a systems-based approach for osmotic studies.