SQLCODE +802: Arithmetic exception (warning)

What it means

Arithmetic in SQL can fail. A sum can overflow its data type, a decimal result can lose digits, and a division by zero has no answer. Where the expression sits decides what Db2 does.

• If the failing expression is in the SELECT list of the outer SELECT and has an indicator variable, Db2 returns +802. That expression comes back with an indicator of -2, and every other column is returned as normal.
• Anywhere else, such as a WHERE clause, a SET clause or an aggregate, or without an indicator variable, the statement fails with -802. Its SQLSTATE is 22012 for a zero divide and 22003 for the rest.

The message names the exception type, the operation and the data type. Exception types include DECIMAL OVERFLOW, DIVIDE EXCEPTION, EXPONENT OVERFLOW, FIXED POINT OVERFLOW, OVERFLOW, UNDERFLOW, OUT OF RANGE and ZERO DIVIDE. Operations are addition, subtraction, multiplication, division, negation or a built-in function. The data type shown can be the type of Db2's internal copy rather than the column's own type. Division by zero on DECFLOAT data doesn't raise this exception.

On Db2 for i, SQL0802 means "Data conversion or data mapping error" and can come with +802 or -802. It has its own list of error types, so read the cause text in the job log rather than applying the z/OS meaning.

Common causes

• A division where the divisor column is zero for some rows, such as AMOUNT / QUANTITY.
• An addition or multiplication in the SELECT list that grows past the range of INTEGER or the precision of a DECIMAL result.
• Multiplying two DECIMAL columns, which can give a result with more digits than Db2 can carry.
• An expression on a SMALLINT or INTEGER column that overflows on large values.
• Bad data, such as a huge or zero value that shouldn't be in the column at all.

How to find the problem

1. Read the exception type, operation, data type and position number. The position is the expression's place in the SELECT list.
2. Find rows where that expression's indicator came back as -2. Those rows have the values that caused it.
3. For zero divide, query the divisor column for zeros. For overflow, look at the largest values involved.
4. Check the ranges of the numeric data types involved against the values in the table.

How to fix it

• Guard divisions with a CASE expression, for example CASE WHEN QUANTITY = 0 THEN NULL ELSE AMOUNT / QUANTITY END.
• Cast to a wider type before the arithmetic, such as BIGINT, or DECIMAL with more digits.
• Fix bad data at its source if zero or huge values shouldn't exist.
• Keep indicator variables on every expression in the SELECT list, and treat -2 as an error in the program.

Technical reviewer: Albert Joseph, WND Software.

Last reviewed 2026-10-18 by the WND Software team. Written in our own words. For the full reference, check IBM’s Db2 for z/OS documentation for your release.