AbstractOils and other surface films used against mosquito larvae may act by flooding the tracheal system with oil, by disrupting the surface forces that allow larvae to rest at the surface, by toxicity, or by eliciting chemosensory responses. In an attempt to identify diagnostic symptoms of these modes of action we treated fourth‐stage larvae of Culex pipiens L. form molestus Forskål (Diptera: Culicidae: Culicinae) with agents operating in a single mode (Ondina oil for flooding with oil, nicotine for toxicity, and the detergent Triton X‐100 for disruption of surface forces), in two modes (silicone oil) or in three (eucalyptus oil, citronellal, or caproic acid ethyl ester). We monitored the time course of flooding and immobilisation, used experiments to separate volatile toxicity from toxicity of the aqueous solution, and used video and The Observer software to analyse larval behaviour.Larvae that experienced tracheal flooding applied their mouthparts to the siphon (tail nibbling), often losing contact with the surface while doing so and falling to the bottom. Nicotine immobilised larvae without interfering with surface forces, and the larvae remained immobile hanging from the air‐water interface for long periods. In detergent, larvae made repeated unsuccessful attempts to thrust the siphon through the surface. The behaviour‐modifying chemicals caproic acid ethyl ester, eucalyptus oil and citronellal all operated in more than one mode. They all decreased the proportion of time larvae spent at the surface, whereas Ondina oil and silicone fluid increased it.Using this approach it should be possible to identify the modes of action of novel larvicidal agents. This could form a basis for rational design of larvicides giving an optimal compromise between efficacy against mosquito larvae and minimal damage to non‐target organisms.