Internal agency records of the Internal Revenue Service (IRS) were surveyed to determine the availability of statistical data on Federal tax violations. How such data might be combined to measure the extent of tax violations was examined, along with their distribution and changing character over time. Three direct and three indirect measurement techniques for estimating offense prevalence are examined: (1) the 'random investigation' method, (2) self-reports, (3) cross-validation matching third-party reports with self-report data, (4) criterion-based predictive formulas from tax data, (5) noncriterion-based estimates from monetary data, and (6) residual estimators based on difference between national income and tax series. Special attention is given to the importance and difficulty of separating criminal and other serious tax violations from general violations. Problems created by statutory law and its changing requirements are also considered. The report concludes with an examination of available information on enforcement activities and sanctions from the IRS's management information systems. Problems of data reliability and the difficulties of matching information across separate data systems are assessed. Criminal and civil penalty statutes enforced by the IRS, the coverage and evolution of major tax-penalty provisions, historical statistics on sources of Federal revenues, and detailed categories coded by the IRS on criminal investigations cases are appended. Tabular and graphic data, footnotes, and approximately 270 references are provided. (Author abstract modified).