To find the reduced row echelon form (RREF) of a matrix, use row operations to give every nonzero row a leading 1, move those leading 1s to the right as you go down, and clear every other entry in their columns. This method is called Gauss-Jordan elimination. Below are the steps, a short example, a full 3×4 system, and the special cases you will meet.
The Gauss-Jordan steps
- Find the pivot column. Start at the leftmost column that is not all zeros.
- Get a nonzero pivot. If the entry in the current row is 0, swap in a lower row that has a nonzero entry in this column.
- Make the pivot 1. Multiply the row by 1 divided by the pivot.
- Clear below the pivot. Add a multiple of the pivot row to each row below so the column has zeros there.
- Move down and right. Repeat steps 1–4 for the next row, ignoring the rows already used. Stop when you run out of rows or columns. The matrix is now in row echelon form.
- Clear above the pivots. Working from the last pivot back to the first, add multiples of each pivot row to the rows above it until each pivot is the only nonzero entry in its column.
Steps 1–5 are the forward phase (Gaussian elimination) and step 6 is the backward phase. You can also clear above and below each pivot as soon as you create it, which is what the RREF calculator does. Both orders give the same answer, because every matrix has exactly one RREF.
Only three row operations are allowed, and none of them changes the solutions of a system:
- swap two rows;
- multiply a row by a nonzero number;
- add a multiple of one row to another row.
Example 1: a 2×3 matrix in two steps
| 1 | 2 | 3 |
| 2 | 5 | 8 |
The first pivot is already 1, so step 4 comes first: subtract twice row 1 from row 2 to clear the entry below it.
| 1 | 2 | 3 |
| 0 | 1 | 2 |
The second pivot is also 1. This is row echelon form. Now clear above the second pivot by subtracting twice row 2 from row 1.
| 1 | 0 | −1 |
| 0 | 1 | 2 |
Both pivot columns contain a single 1 and zeros elsewhere. The matrix has rank 2, and its third column is a non-pivot column.
Example 2: a 3×4 system that needs a row swap
Solve the system y + 2z = 3, x + z = 2, 2x + y = 1. Write it as an augmented matrix, with the constants in the last column:
| 0 | 1 | 2 | 3 |
| 1 | 0 | 1 | 2 |
| 2 | 1 | 0 | 1 |
The top-left entry is 0, so it cannot be a pivot. Each step below says why it is done; the shaded row is the one that changed and the outlined entry is the pivot being used.
Column 1 has a 0 in the pivot position of row 1, so swap in row 2, which has a nonzero entry there.
| 1 | 0 | 1 | 2 |
| 0 | 1 | 2 | 3 |
| 2 | 1 | 0 | 1 |
Clear the entry below the pivot: add −2 × row 1 to row 3 so column 1 becomes 0 there.
| 1 | 0 | 1 | 2 |
| 0 | 1 | 2 | 3 |
| 0 | 1 | −2 | −3 |
Clear the entry below the pivot: add −1 × row 2 to row 3 so column 2 becomes 0 there.
| 1 | 0 | 1 | 2 |
| 0 | 1 | 2 | 3 |
| 0 | 0 | −4 | −6 |
Multiply row 3 by −1/4 so its pivot in column 3 becomes 1.
| 1 | 0 | 1 | 2 |
| 0 | 1 | 2 | 3 |
| 0 | 0 | 1 | 3/2 |
Clear the entry above the pivot: add −1 × row 3 to row 1 so column 3 becomes 0 there. Clearing above pivots is what turns REF into RREF.
| 1 | 0 | 0 | 1/2 |
| 0 | 1 | 2 | 3 |
| 0 | 0 | 1 | 3/2 |
Clear the entry above the pivot: add −2 × row 3 to row 2 so column 3 becomes 0 there. Clearing above pivots is what turns REF into RREF.
| 1 | 0 | 0 | 1/2 |
| 0 | 1 | 0 | 0 |
| 0 | 0 | 1 | 3/2 |
The coefficient part is now the identity matrix, so the system has exactly one solution. Read it from the last column: x = 1/2, y = 0, z = 3/2. Substitute back to check: 0 + 2·3/2 = 3, 1/2 + 3/2 = 2, and 2·1/2 + 0 = 1.
Special cases you will meet
A row of zeros
If a row becomes all zeros, that equation was a combination of the others. Move it to the bottom and carry on. A square matrix with a zero row in its RREF is not invertible.
No solution
If a row reads 0 0 0 | 1, it says 0 = 1, so the system is inconsistent. For x + y + z = 2, x + 2y + 3z = 5 and 2x + 3y + 4z = 9, the third equation contradicts the sum of the first two:
| 1 | 0 | −1 | 0 |
| 0 | 1 | 2 | 0 |
| 0 | 0 | 0 | 1 |
Free variables and infinitely many solutions
If a coefficient column has no pivot and no row reads 0 = 1, that variable is free and the system has infinitely many solutions. Here x₃ has no pivot, so it becomes a parameter t:
| 1 | 0 | 1 | 5/3 |
| 0 | 1 | −1 | 2/3 |
| 0 | 0 | 0 | 0 |
- x1 = 5/3 − 1·t1
- x2 = 2/3 + 1·t1
- x3 = t1
The augmented matrix calculator classifies every system this way and writes the parametric solution for you.
What makes a matrix RREF?
- Any rows made entirely of zeros are at the bottom.
- The first nonzero entry in each nonzero row, its pivot, is 1.
- Each pivot sits to the right of the pivot in the row above.
- Every other entry in a pivot column is 0.
A matrix that meets only the first and third conditions is in row echelon form. Compare the two in REF vs. RREF. The number of pivots is the rank of the matrix, and because the RREF is unique, so are the pivot positions.
Common mistakes
- Changing only part of a row. Every operation applies to the whole row, including the constants column.
- Multiplying a row by 0. That destroys information and is not a valid row operation.
- Stopping at row echelon form. RREF also needs zeros above each pivot.
- Using a row that already has a pivot to create a new pivot. Work downward; earlier pivot rows are only used for clearing.
- Rounding early. Keep fractions such as 1/3 exact until the end; decimals like 0.333 introduce errors that can hide a zero.
- Sign slips. When a factor is negative, write it in brackets, for example R3 ← R3 + (−2)R1.
Practice problems
Reduce each matrix by hand, then open the answer. You can also check your working step by step in the RREF calculator.
Reduce this 2×2 matrix.
Practice problem 1 1 3 2 7 Show the answer
Answer to practice problem 1 1 0 0 1 Two pivots, so the RREF is the identity: the matrix is invertible and has rank 2.
Reduce this 2×3 matrix. Start by scaling row 1.
Practice problem 2 2 4 6 1 3 5 Show the answer
Answer to practice problem 2 1 0 −1 0 1 2 Rank 2. Column 3 has no pivot, so it is a combination of the first two columns.
Reduce this 3×4 augmented matrix and describe the solutions.
Practice problem 3 1 −1 2 1 2 0 2 4 0 3 −3 3 Show the answer
Answer to practice problem 3 1 0 1 2 0 1 −1 1 0 0 0 0 The last row is all zeros and x₃ has no pivot, so the system has infinitely many solutions: x₁ = 2 − t, x₂ = 1 + t, x₃ = t.
Questions about finding RREF
Is the RREF of a matrix unique?
Yes. Whatever row operations you choose, and in whatever order, a matrix has exactly one reduced row echelon form. Row echelon form is not unique.
Do I have to follow the steps in a fixed order?
No. Any sequence of valid row operations that reaches a matrix meeting the four RREF rules gives the same result. A fixed order, such as the Gauss-Jordan steps above, just makes the work systematic and easier to check.
How do I find the rank from the RREF?
Count the pivots, which is the same as counting the nonzero rows. The example 2×3 matrix above has rank 2.
What is the difference between Gaussian and Gauss-Jordan elimination?
Gaussian elimination clears only below each pivot and stops at row echelon form, then solves by back substitution. Gauss-Jordan elimination also clears above each pivot and ends at RREF, where the solution can be read directly. See REF vs. RREF.
Can a graphing calculator find RREF?
Yes. The TI-83 Plus and TI-84 Plus have an rref( command in the matrix menu, but they show only the final matrix, not the steps. See how to find RREF on a TI-84.
Reference: University of Texas at Austin, Matrix Operations appendix. Examples independently calculated with exact arithmetic. Sources checked September 25, 2026.