Random experiments could allow the government to test tax policies before they are enacted into general law. Such experiments can be revenue-neutral, with the tax authority ensuring ex post that average tax revenues received from taxpayers in the treatment and control groups are equal. Taxpayers might thus volunteer even for experiments that would broaden the tax base, for example by eliminating deductions. Continued participation by taxpayers in such experiments would indicate that the proposed reforms are efficient at least if externalities are disregarded. Non-revenue-neutral experiments raise greater concerns about horizontal inequity, but they may be helpful in addressing questions about effects of tax rates and in increasing participation.