We design a series of laboratory experiments to investigate the effects of purchasing insurance and of pre-filled claim forms on dishonesty in loss reporting. In our experiment, participants report the outcome of privately rolling two dice where the numbers rolled map to a payoff distribution with the possibility of losses in earned income. Prior to this reporting task, participants bid for a limited number of insurance contracts which issue an indemnity payment equal to each insured individual’s reported loss. We find that dishonest reporting is significantly more prevalent among insured individuals relative to the uninsured, consistent with an ‘entitlement bias’. Further we find that prefilling the reporting form with the most probable outcome only modestly constrains dishonest reporting among both insured and uninsured individuals. We explore reasons why pre-filled forms should be applied with caution.