Citizen science (CS) initiatives can transform how some ecological data are collected. Herbivory is a fundamental ecological interaction, but herbivory rates in many natural systems are unknown due in part to lack of personnel for monitoring efforts. This limits our ability to understand broad ecological patterns relevant to herbivory. Fortunately, accurate and reliable visual estimation techniques for assessing herbivory could be amenable to CS approaches. In 2008, I developed a CS training initiative (the Million Leaf Project, MLP) to measure herbivory based on a seven‐category visual assessment of leaf area removed (LAR). From 2010 to 2018, 394 citizen scientists assessed damage on 175,421 leaves to test the hypothesis that herbivory varies between understory and canopy leaves in a Peruvian tropical forest. In support of this hypothesis, the longitudinal CS data reveal that understory leaves consistently experience more herbivory than do canopy leaves on average (18.3% vs. 12.3%, P < 0.001), a difference that was consistent regardless of CS observer age. Furthermore, data integrity was high, even though younger participants showed some leaf selection bias. The MLP is based on a simple technique, intended to broaden public participation in ecological science, and applicable to any ecological system in which herbivory or leaf damage occurs.