A tiered protocol for assessing ecological risks has been applied to the rice pesticide profoxydim. The initial assessment (Tier I) was based on toxicity exposure ratio (TER) calculations based on laboratory data using a worst-case rice scenario. The first refinement (Tier II) was based on direct toxicity assessment (DTA) of water samples collected during a field-mesocosm study. Finally, a higher-tier assessment on the in situ assessment of paddy community responses (field-mesocosm-Tier III) was performed. A successive application of three pesticides, the herbicides azimsulfuron, propanil and the insecticide malathion, was used as reference controls. The refined assessments indicated a lower risk than that predicted from TER estimations. DTA-based Tier II showed toxicity effects only for concentrations above the recommended dose of profoxydim. Effects for reference controls were observed in DTA which were not expected from Tier I. The field-mesocosm study confirmed these effects but also showed that they were transient and of low relevance.