Many recent ecotoxicological studies suggest a relationship between freshwater contamination and increasing abundances of motile diatoms (potentially able to move). The capacity to escape would present advantages to species in polluted environments. However, actual motility as a response to toxicants had not been described and required experimental validation. We designed a specific experiment to assess how a field-isolated diatom (Gomphonema gracile) distributes energy to in situ resistance (increased population growth or photosynthesis) and escape (behavioral changes), when exposed to increasing concentrations of the herbicide metolachlor. We report here the dose-time dependent responses of G. gracile populations. They coped with low contamination by resisting in situ, with early hormetic responses highlighted by stimulation of chlorophyll-a fluorescence. At a higher dose, harmful impacts were observed on growth after a few days, but an earlier behavioral response suggested that higher motility (percentage of motile individuals and mean distance crossed) could be involved in escape. Our findings bring new arguments to support the implementation of real measurements instead of motility traits in toxicity assessment. Specifically, motion descriptors have been used as early-warning indicators of contamination in our study. Further works should address the reliability of these endpoints in more complex conditions (interspecific variability, behavior in the field).