We investigate experimentally whether individuals or groups are more lied to, and how lying depends on the group size and the monetary loss inflicted by the lie. We employ an observed cheating game, where an individual's misreport of a privately observed number can monetarily benefit her while causing a loss to either a single individual, a group of two or a group of five. As the privately observed number is known to the experimenter, the game allows to study both, whether the report deviates from the observed number and also by how much. Treatments either vary the individual loss caused by a given lie (keeping the total loss constant), or the total loss (keeping the individual loss constant). We find more lies toward individuals than toward groups. Liars impose a larger loss with their lie when that loss is split among group members rather than borne individually. The size of the group does not affect lying behavior.