Worldwide, large-scale biodiversity monitoring schemes are developing and involve many non-specialist volunteers. If the opening of schemes to non-specialists allows for the gathering of huge amounts of data, their quality represents a controversial issue. In the framework of the French Garden Butterfly Observatory (FGBO), we studied non-specialist volunteer identification errors based on identifications provided during a one-shot experiment. With 3492 butterfly pictures sent by 554 non-specialist volunteers, we directly measured identification errors and misidentification rates for each butterfly species or species group targeted by the FGBO. The results showed that when non-specialist volunteers identified butterflies at the species level, identification errors (i.e., the misidentification rate) reached 20.9%. It was only 5.0% when FGBO species groups were used. This study provides novel insights into the trade-off between data quantity and quality provided by non-specialist volunteers and shows that if protocols, research questions and identification levels are adapted, participatory monitoring schemes relying on non-specialists represent a powerful and reliable tool to study common species at a large scale and on a long-term basis.