In Cambodia, despite an impressive decline in prevalence over the last 10 years, malaria is still a public health problem in some parts of the country. This is partly due to vectors that bite early and outdoors reducing the effectiveness of measures such as Long-Lasting Insecticidal Nets. Repellents have been suggested as an additional control measure in such settings. As part of a cluster-randomized trial on the effectiveness of topical repellents in controlling malaria infections at community level, a mixed-methods study assessed user rates and determinants of use. Repellents were made widely available and Picaridin repellent reduced 97% of mosquito bites. However, despite high acceptability, daily use was observed to be low (8%) and did not correspond to the reported use in surveys (around 70%). The levels of use aimed for by the trial were never reached as the population used it variably across place (forest, farms and villages) and time (seasons), or in alternative applications (spraying on insects, on bed nets, etc.). These findings show the key role of human behavior in the effectiveness of malaria preventive measures, questioning whether malaria in low endemic settings can be reduced substantially by introducing measures without researching and optimizing community involvement strategies.