Using RREF to Solve a 3x3 System of Equations

Three equations, three unknowns. That combination shows up constantly in physics, engineering, and economics, and RREF gives you a clean, mechanical way to crack it every time. Once you understand how augmented matrices work, the 3x3 case is a natural extension of the 2x2 process.

Setting Up the Problem

Start with a concrete system:

2x +  y - z =  8
-3x - y + 2z = -11
-2x +  y + 2z = -3

The goal is to find the unique values of x, y, and z that satisfy all three equations simultaneously. RREF turns that goal into a row-reduction exercise rather than a substitution puzzle.

Write the augmented matrix by pulling the coefficients and the constants:

[  2   1  -1 |   8 ]
[ -3  -1   2 | -11 ]
[ -2   1   2 |  -3 ]

The vertical bar just separates the coefficient side from the constant side. Everything to the left of it comes from the x, y, z columns; everything to the right comes from the right-hand side of each equation.

Executing the Row Reduction

Step 1: Create a leading 1 in row 1

Divide R1 by 2:

[  1   1/2  -1/2 |   4 ]
[ -3  -1    2    | -11 ]
[ -2   1    2    |  -3 ]

Step 2: Eliminate the first column below the pivot

Use R2 = R2 + 3·R1 and R3 = R3 + 2·R1:

R2: -3 + 3(1) = 0,  -1 + 3(1/2) = 1/2,  2 + 3(-1/2) = 1/2,  -11 + 3(4) = 1
R3: -2 + 2(1) = 0,   1 + 2(1/2) = 2,    2 + 2(-1/2) = 1,    -3 + 2(4) = 5

Matrix after this step:

[  1   1/2  -1/2 |  4 ]
[  0   1/2   1/2 |  1 ]
[  0   2     1   |  5 ]

Step 3: Create a leading 1 in row 2

Multiply R2 by 2:

[  1   1/2  -1/2 |  4 ]
[  0   1     1   |  2 ]
[  0   2     1   |  5 ]

Step 4: Eliminate column 2 above and below the pivot

Use R1 = R1 - (1/2)·R2 and R3 = R3 - 2·R2:

R1: 1 - 0 = 1,  1/2 - 1/2 = 0,  -1/2 - 1/2 = -1,  4 - 1 = 3
R3: 0 - 0 = 0,    2 - 2 = 0,       1 - 2 = -1,      5 - 4 = 1

[  1   0  -1 |  3 ]
[  0   1   1 |  2 ]
[  0   0  -1 |  1 ]

Step 5: Create a leading 1 in row 3

Multiply R3 by -1:

[  1   0  -1 |  3 ]
[  0   1   1 |  2 ]
[  0   0   1 | -1 ]

Step 6: Eliminate column 3 above the pivot

Use R1 = R1 + R3 and R2 = R2 - R3:

R1: 1, 0, 0 | 3 + (-1) = 2
R2: 0, 1, 0 | 2 - (-1) = 3

Final RREF:

[  1   0   0 |  2 ]
[  0   1   0 |  3 ]
[  0   0   1 | -1 ]

Reading the Solution

An identity matrix on the left side means the system has a unique solution. Each row directly states one variable:

Verify by substituting back into the original system:

2(2) + 3 - (-1) = 4 + 3 + 1 = 8  ✓
-3(2) - 3 + 2(-1) = -6 - 3 - 2 = -11  ✓
-2(2) + 3 + 2(-1) = -4 + 3 - 2 = -3  ✓

All three check out. The full RREF process guarantees this kind of clean readout whenever a unique solution exists.

When the System Doesn't Have a Unique Solution

Not every 3x3 system lands on a nice identity matrix. Two other outcomes are possible: infinitely many solutions or no solution at all.

Dependent system (infinitely many solutions)

Suppose row reduction produces a zero row:

[  1   0   2 |  4 ]
[  0   1  -1 |  3 ]
[  0   0   0 |  0 ]

The third row says 0 = 0, which is always true but gives no information. This means one variable (here, z) is free. Let z = t for any real number t, then:

There are infinitely many solution triples, one for every value of t.

Inconsistent system (no solution)

If a zero row produces a nonzero constant:

[  1   0   2 |  4 ]
[  0   1  -1 |  3 ]
[  0   0   0 |  7 ]

Row 3 reads 0x + 0y + 0z = 7, which is impossible. The system is inconsistent and has no solution. The full discussion of consistent vs inconsistent systems covers what these cases mean geometrically.

Connecting RREF to Broader Problem-Solving

The row operations here follow the same three types used in every RREF computation: swapping rows, multiplying a row by a nonzero scalar, and adding a multiple of one row to another. None of those operations change the solution set, which is why the final RREF carries the same answer as the original system.

For larger systems (4x4, 5x5, and beyond), the algorithm scales without any new ideas. The bookkeeping gets heavier, which is exactly why using RREF to solve linear systems is worth mastering with a calculator for anything beyond 3x3.

One practical tip: track fractions carefully in the early steps. A sign error on a coefficient in step 2 propagates through every subsequent row, making the final answer look bizarre rather than obviously wrong. Doing one substitution-check at the end, as shown above, catches that kind of mistake before it causes real trouble.

Frequently Asked Questions

How many row operations does a 3x3 system typically require?

For a 3x3 system with a unique solution, expect roughly 8 to 12 individual row operations to reach full RREF. The exact count varies depending on the coefficients. Systems with large or inconvenient entries sometimes require a row swap early on to avoid working with awkward fractions.

What does it mean if two rows become identical during reduction?

If two rows reduce to identical entries, the matrix is rank-deficient. The duplicate row becomes a zero row after subtraction, which signals either a dependent system (infinitely many solutions) or, combined with a nonzero constant, an inconsistent system. Either way, a unique solution no longer exists.

Can I use a different pivot order?

Yes. The standard algorithm works left to right, top to bottom, but partial pivoting (choosing the row with the largest absolute value in the pivot column) can reduce rounding errors in hand calculations and is especially useful in software implementations. The final RREF is always the same regardless of pivot order, provided no arithmetic mistakes are made.

Is RREF the same as Gaussian elimination?

Not exactly. Gaussian elimination produces row echelon form (REF), where the entries above and below each pivot are not necessarily zero. RREF goes further: it eliminates entries both above and below each pivot and scales every pivot to 1. REF is enough to solve by back-substitution, but RREF lets you read the solution directly from the matrix with no extra work.