IRS Service Center Notices Issued
For Math Errors and Under-Reporting
Adjustments From Computerized
Document Matching

Fiscal Year Returns Filed*
Math Error
Returns           Percent
Under-Reporting
 Returns            Percent
1981 93,052,000 na   1,200,000 1.3
1982 94,013,000 na   2,900,000 3.1
1983 95,419,000 na   2,900,000 3.0
1984 95,541,300 na   3,900,000 4.1
1985 96,496,900 na   3,600,000 3.7
1986 99,529,000 na   3,200,000 3.2
1987 101,750,800 na   2,242,000 2.2
1988 103,251,000 na   3,800,000 3.7
1989 107,029,000 na   3,650,000 3.4
1990 109,868,400 na   2,950,000 2.7
1991 112,304,900 na   4,840,000 4.3
1992 113,829,200 4,985,000 4.38 3,771,509 3.3
1993 114,718,900 4,088,000 3.56 2,723,830 2.4
1994 113,754,400 4,059,000 3.57 2,645,075 2.3
1995 114,683,400 6,102,000 5.32 2,711,086 2.4
1996 116,059,700 4,751,000 4.1 1,930,326 1.7
1997 118,362,600 *5,984,000 5.1 931,354 0.8
1998 120,342,400 5,669,000 4.7 1,726,098 1.4
1999 121,829,470 6,552,000 5.4 1,770,695 1.5
2000 124,887,140 5,751,000 4.6 1,353,545 1.1
2001 127,097,210 6,082,967 4.8 1,161,901 0.9
2002 129,444,947 **13,315,765 **10.3 1,491,139 1.2
2003 130,341,159 ***4,967,703 3.8 1,561,068 1.2
2004 130,134,277 ***7,342,484 5.6 ****1,948,363 1.5
2005 130,576,852 3,163,794 2.4 2,837,000 2.2
2006 132,275,830 3,159,077 2.4 3,209,000 2.4


Source: IRS March 26, 2001 Submission to Senate Finance Committee; September 2002 submission to IRS Oversight Board; "Report to Congress: IRS Tax Compliance Activities," July 15, 2003, Internal Revenue Service, Department of the Treasury; FY 2002-2003 IRS Data Books. Return counts represent prior calendar year filings based upon internal IRS reports.

*Expanded math error authority was implemented in calendar year 1997 for tax year 1996; data prior to FY 1998 does not include earned income tax credit (EITC) or taxpayer identification number (TIN) "math errors." Return counts are for prior calendar year.
**Spike caused by taxpayer confusion with the advance payment of the rate reduction credit.
***Basis for counts revised between 2003 and 2004 to exclude prior year returns. If 2003 had been reported using same definition, would have been 4,024,981. Spike in 2004 largely due to large jump in child care credit math errors. Figures for 2005 and 2006 now count notices rater than math errors since a single return may have more than one error. If rate had been based on math errors in 2006, it would have been 3.1, instead of 2.4.
****Basis of counts changed from number of contacts to cases closed in 2004 and later years.

Note that "na" means not available.


Transactional Records Access Clearinghouse, Syracuse University
Copyright, 2003, 2004, 2005