Consistent vs. Inconsistent Systems: How RREF Tells You

Row reduction is one of the few algebraic procedures that gives you a definitive verdict. Feed any linear system into the RREF algorithm and the result will tell you, plainly and without ambiguity, whether solutions exist at all. That verdict hinges on one key distinction: consistent versus inconsistent.

What These Terms Actually Mean

A consistent system has at least one solution. That solution might be unique (a single point where all equations intersect) or there might be infinitely many (a line, plane, or higher-dimensional solution set). Either way, there exists at least one assignment of values to the variables that satisfies every equation simultaneously.

An inconsistent system has no solution at all. The equations contradict each other. No matter what values you plug in, at least two equations will conflict. Geometrically, think of two parallel lines in the plane: they never meet.

The practical problem is that inconsistency is not always obvious from the original equations. A system of five equations in four unknowns might look fine until you work through it and hit a contradiction on the third row. RREF makes that contradiction explicit, which is why it is the standard diagnostic tool.

The No-Solution Row: What It Looks Like

After row reducing the augmented matrix of a system, scan each row. The signal for inconsistency is a row of this form:

[ 0  0  0  ...  0 | c ]    where c ≠ 0

Every entry to the left of the vertical bar is zero. The entry to the right is nonzero. That row translates directly to the equation:

0·x₁ + 0·x₂ + ... + 0·xₙ = c

Which simplifies to 0 = c. Since c ≠ 0, this is a false statement. No values of x₁ through xₙ can ever make it true, so the system has no solution.

If you see that row anywhere in the RREF, stop. The system is inconsistent and the rest of the matrix is irrelevant to whether solutions exist.

The Rank Test

There is a more compact way to state the same thing using rank. The rank of a matrix is the number of nonzero rows in its RREF (equivalently, the number of pivot positions).

For a system with coefficient matrix A and augmented matrix [A|b]:

Consistent if and only if r = r*

Inconsistent if and only if r* = r + 1

The augmented matrix can only have rank equal to r or r + 1. If adding the constant column b increases the rank, it means b introduced a new pivot, which can only happen in that telltale zero-row position. That extra pivot is exactly the 0 = c contradiction.

For more on computing rank directly, see how to find the rank of a matrix.

Worked Example: An Inconsistent System

Consider this 3x2 system:

x + 2y = 5
3x + 6y = 9
2x + 4y = 10

Form the augmented matrix:

[ 1  2 | 5  ]
[ 3  6 | 9  ]
[ 2  4 | 10 ]

Row reduce. Subtract 3 times row 1 from row 2, and 2 times row 1 from row 3:

[ 1  2 | 5  ]
[ 0  0 | -6 ]
[ 0  0 | 0  ]

Row 2 reads 0x + 0y = -6, which is 0 = -6. Inconsistent. The rank of the coefficient matrix is 1 (one pivot in column 1). The rank of the augmented matrix is 2 (a second pivot appears in the last column of row 2). Since 2 = 1 + 1, the rank test confirms no solution.

The original three equations look superficially different, but the second line is just 3 times the first with a different right-hand side. That mismatch is the contradiction.

Worked Example: A Consistent System

Now take:

x + 2y - z = 3
2x + 5y + z = 8
x + 3y + 2z = 5

Augmented matrix:

[ 1  2  -1 | 3 ]
[ 2  5   1 | 8 ]
[ 1  3   2 | 5 ]

R2 = R2 - 2·R1, R3 = R3 - R1:

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

R3 = R3 - R2:

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

Row 3 is all zeros including the right side, so it contributes no contradiction. Back-substitute from row 2: y + 3z = 2, so y = 2 - 3z. Then from row 1: x = 3 - 2y + z = 3 - 2(2 - 3z) + z = -1 + 7z.

The system is consistent with infinitely many solutions: x = -1 + 7t, y = 2 - 3t, z = t for any real t. One free variable, one parameter. See no solution, one, or infinitely many for a full treatment of free variables.

Connecting This to Augmented Matrices

The augmented matrix is the right tool here precisely because it carries the constants b alongside the coefficients. If you only row reduce the coefficient matrix A, you lose the information needed to detect inconsistency. A pivot appearing in the augmented column (the (n+1)-th column) is the mechanism by which RREF announces a contradiction.

This is also why you should understand what the augmented matrix encodes before running RREF on it. A clear explanation is in augmented matrices explained.

For a walkthrough of how to actually solve a consistent system once RREF gives you a unique or parametric solution, see using RREF to solve linear systems.

Frequently asked questions

Can a system be inconsistent if it has more unknowns than equations?

Yes. The count of equations versus unknowns does not determine consistency on its own. An underdetermined system (more unknowns than equations) is often consistent with infinitely many solutions, but it can still be inconsistent if the equations contradict each other. For example, x + y = 3 and x + y = 7 is inconsistent despite having more unknowns than equations, because the left sides are identical but the right sides differ.

What if every row in the RREF has all zeros?

If every row reduces to [0 0 ... 0 | 0], the augmented matrix has rank zero and the system is trivially consistent. Every variable is free. This happens when all original equations are scalar multiples of each other and the right-hand sides match proportionally. The solution set is the entire space spanned by the free variables.

Does RREF always produce a unique answer for consistency?

Yes, RREF is unique for any given matrix. Unlike partial row echelon form, which can vary depending on the sequence of operations, the fully reduced form is canonical. That means two people row reducing the same augmented matrix by different sequences of valid operations will always arrive at the same RREF, and therefore the same verdict on consistency.

Can the no-solution row appear in the middle of the RREF?

In a proper RREF, pivot rows appear at the top and zero rows sink to the bottom. So in the final reduced form, a no-solution row would be among the bottom rows. During intermediate steps, a contradiction row might appear anywhere, but after full reduction, row ordering places it at the bottom. The position within the RREF does not change the conclusion: any row with all-zero coefficients and a nonzero constant means the system is inconsistent.