Cooperative interactions offer the inherent possibility of cheating by each of the interacting partners. A key challenge to behavioural observers is to recognize these conflicts, and find means to measure reliably cheating in natural interactions. Cleanerfish Labroides dimidiatus cheat by taking scales and mucus from their fish clients and such dishonest cleaning has been previously recognized in the form of whole-body jolts by clients in response to cleaner mouth contact. In this study, we test whether jolts may be a general client response to cheating by cleaners. We experimentally varied the ectoparasite loads of yellowtail damselfish (Microspathodon chrysurus), a common client of the cleaning goby Elacantinus evelynae, and compared the rates of jolts on parasitized and deparasitized clients. As predicted if jolts represent cleaner cheating, deparasitized clients jolted more often than parasitized clients, and overall jolt rates increased over time as client parasite load was presumably reduced by cleaning activity. Yellowtail damselfish in the wild jolted significantly less frequently than those in captivity, which is consistent with a loss of ectoparasites during capture. Our results suggest that jolts by clients of cleaning gobies are not related to the removal of ectoparasites. Client jolts may therefore be a generally accurate measure of cheating by cleanerfish.