||High-throughput screens typically search large libraries of chemical or genetic perturbants for new drugs or genetic regulators. As automated microscopy and image analysis has improved, screens based on cultured cells have proven to be powerful and versatile. However, many diseases and biological pathways can be better studied in whole animals, particularly diseases that involve organ systems and multicellular interactions, such as metabolism and infection. The roundworm Caenorhabditis elegans is a well studied and effective model system for such biological processes, and recent advances in robotic sample preparation have made it possible to use C. elegans in high-throughput screens. High-throughput screens are usually limited to low-resolution images without time-lapse information, and the number of images for a standard screen often exceeds 100 000, ruling out visual inspection. Each image (i.e., well) of a screen typically holds 15 worms, and apart from robustness to noise and variations in illumination, the largest challenge from an image analysis point of view is clustering of worms, which prevents extraction of descriptive features from individual animals. Apart from presenting some screening results based on feature extraction from clustered worms, I will also present a novel approach for the extraction of individual C. elegans from clusters. Our approach is based on a graph search combined with a low-dimensional shape-descriptor space defining shape probability. Once individual worms are well defined, features such as shape and color can be extracted to discriminate between, e.g., live and dead worms, and worms can be digitally straightened and aligned in order to compare staining patterns of biologically relevant markers.