Elimination solves a system by adding or subtracting scaled equations so one variable cancels, leaving a single-variable equation you can solve. In a two-variable system, the reliable order is: multiply one or both equations to match a coefficient, add or subtract to eliminate, solve for one variable, then back-substitute. The condition that matters is the multiplier: if neither equation can be scaled to produce matching or opposite coefficients for the same variable, add-and-subtract alone won't work and you need the special-case analysis below.

What Elimination Actually Does to Each Equation

Elimination rests on one property: multiplying both sides of an equation by the same nonzero number produces an equivalent equation. That's why you can scale 2x + 3y = 12 by 3 and still trust the result. The solution set doesn't change, only the appearance of the line or plane.

Adding two equations is a second equivalence. If A = B and C = D, then A + C = B + D. In geometric terms, adding two linear equations produces a third line that passes through every point both original lines share. The new line is a "combination" of the originals. Here's the key: that combination line contains the intersection point, but it isn't the same line, so you've replaced one constraint with a simpler one without losing the answer.

The variable you target should be whichever has coefficients that are already the same, already opposites, or easily scaled into agreement. Look at the coefficients before you touch anything. If x has coefficients 2 and 3, multiply the first equation by 3 and the second by 2 to get 6 and 6 (subtract) or 6 and βˆ’6 (add, if one coefficient was negative). The multiplier is always the other equation's coefficient, swapped, with signs handled by choosing add or subtract.

A worked example makes it concrete. Take 2x + 3y = 12 and 4x βˆ’ 3y = 6. The y-coefficients are 3 and βˆ’3, already opposites. Add the equations: 6x = 18, so x = 3. Substitute back into the first: 2(3) + 3y = 12, so 3y = 6 and y = 2. Solution is (3, 2). Check by plugging into the second equation: 4(3) βˆ’ 3(2) = 12 βˆ’ 6 = 6. Correct.

The Six-Step Procedure, in the Order You Should Run It

Follow this sequence. Skipping step 2 is the most common cause of wasted work.

1: Write both equations in standard form (Ax + By = C). Move variables left and constants right before anything else. 2: Decide which variable to eliminate. Pick the variable whose two coefficients are equal, opposite, or one is a multiple of the other. That saves a multiplication step. 3: Multiply one or both equations so the chosen variable's coefficients are exact opposites. Multipliers are legal only if nonzero. 4: Add the two equations. If coefficients are already opposites, adding is the move. If they're equal but same-sign, subtract one equation from the other. 5: Solve the resulting one-variable equation. This step should feel trivial. If it doesn't, you likely made an arithmetic error in step 3. 6: Back-substitute into whichever original equation looks simplest. Prefer the one with smaller coefficients or fewer negative signs.

Run the procedure on a slightly messier system: 3x + 4y = 10 and 5x βˆ’ 2y = 8. The y-coefficients are 4 and βˆ’2. Multiply the second equation by 2: 10x βˆ’ 4y = 16. Add to the first: 13x = 26, so x = 2. Back-substitute into the first: 3(2) + 4y = 10, so 4y = 4 and y = 1. Solution is (2, 1). Check both originals: 3(2) + 4(1) = 10 βœ“, and 5(2) βˆ’ 2(1) = 8 βœ“.

Failure handling at step 3: if the multipliers would be zero, you've misread the coefficients. Zero has no reciprocal, and y = 0 coefficients only mean that variable is already eliminated, so you can skip directly to step 5. If you get fractions like 7/3 while back-substituting, keep them exact until the final answer. Rounding early is the fastest way to lose a check.

When Multiplication Alone Fails: Inconsistent and Dependent Systems

Sometimes elimination produces a false statement or a tautology. Those outcomes are information, not failure. Three cases exist, and elimination distinguishes them cleanly.

Case 1: unique solution. You get a valid numeric value for the first variable, then a numeric value for the second. Two lines meet at one point.

Case 2: inconsistent (no solution). You get 0 = 5 or a similar contradiction. That means the two lines are parallel and never meet. Example: x + y = 4 and 2x + 2y = 9. Multiply the first by 2: 2x + 2y = 8. Subtract from the second: 0 = 1. False, so no solution. Parallel lines, slopes equal.

Case 3: dependent (infinitely many solutions). You get 0 = 0. The two equations describe the same line. Example: x + y = 4 and 3x + 3y = 12. Multiply the first by 3: same equation. Any (x, y) with x + y = 4 works. The answer is expressed as a parameterized solution: (x, 4 βˆ’ x) for any real x.

Detecting case 2 versus case 3 without fully solving: convert both equations to slope-intercept form (y = mx + b). Same slope, different intercept β†’ inconsistent. Same slope, same intercept β†’ dependent. Same slope and same equation after simplification β†’ all points shared. Different slopes β†’ unique solution, guaranteed.

Three-Variable Systems: Eliminate Down, Then Up

Three-equation, three-variable systems follow the same logic, but you run elimination twice to get from three variables to one. Plan the order before you compute.

Label the equations E1, E2, E3. Pick a target variable, say z. Use E1 and E2 to eliminate z, producing a new equation F1 in (x, y). Use E1 and E3 to eliminate z again, producing F2. Now you have a two-variable system F1, F2 in x and y. Solve it by elimination as before. Back-substitute into E1 to find z. Then check all three originals.

Worked example. E1: x + y + z = 6. E2: 2x βˆ’ y + z = 3. E3: x + 2y βˆ’ z = 2. E1 βˆ’ E2: (x + y + z) βˆ’ (2x βˆ’ y + z) = 6 βˆ’ 3 β†’ βˆ’x + 2y = 3. Call this F1. Add E1 + E3: (x + y + z) + (x + 2y βˆ’ z) = 6 + 2 β†’ 2x + 3y = 8. Call this F2. Solve F1 and F2. Multiply F1 by 2: βˆ’2x + 4y = 6. Add F2: 7y = 14, so y = 2. Back-substitute into F1: βˆ’x + 4 = 3, so x = 1. Substitute into E1: 1 + 2 + z = 6, so z = 3. Check E2: 2(1) βˆ’ 2 + 3 = 3 βœ“. Check E3: 1 + 4 βˆ’ 3 = 2 βœ“. Solution is (1, 2, 3).

The failure mode to watch: if E1 and E2 are multiples of each other (dependent), you'll get 0 = 0 and the system may still have infinitely many solutions depending on E3. With three equations, all three must be consistent. Elimination exposes the conflict if one equation contradicts the pair.

Two-Variable Elimination vs. Substitution: Which to Pick

Both methods solve the same systems. The choice is about efficiency, not correctness.

Situation Elimination wins when… Substitution wins when…
Coefficients of one variable are 1 or βˆ’1 β€” Yes, isolate that variable in one move
Coefficients are already equal or opposite Yes, no scaling needed β€”
Coefficients are 2, 3, 4, or similar small integers Yes, one multiplication step Also fine but slower
Coefficients are fractions like Β½ or β…“ Yes, multiply by denominators first Messy β€” avoid
One equation is already y = (something) or x = (something) β€” Yes, direct insertion
Three-variable system Yes, cleaner to scale Substitution compounds errors

A rule of thumb: elimination shines when you can see a way to match coefficients without much arithmetic. Substitution shines when a variable is already isolated or has coefficient 1. Both are legal on any solvable system, so pick by inspection.

Compare on a real pair. 5x + 3y = 21 and 2x βˆ’ 3y = 7. Elimination adds immediately: 7x = 28, x = 4. Back-substitute into the second: 8 βˆ’ 3y = 7, y = 1/3. Total work: one addition, one substitution. Substitution would require isolating y from the first equation as y = (21 βˆ’ 5x)/3, then substituting into the second β€” more steps, more fractions. Elimination is clearly cleaner here.

Reverse case. y = 3x + 1 and 4x + 2y = 30. Substitution is faster: 4x + 2(3x + 1) = 30 β†’ 10x = 28, x = 2.8, y = 9.4. Elimination requires rewriting the first equation into standard form first. Substitution wins.

Linear Systems vs. Nonlinear Systems: Where the ClichΓ© Fails

Elimination is most associated with linear systems, but it works on some nonlinear ones too, provided one variable still cancels. The skill transfers, with a warning.

Consider xΒ² + yΒ² = 25 and y = xΒ² βˆ’ 5. Substitution is the safer choice here because the second equation gives y in terms of xΒ². Substituting into the first: xΒ² + (xΒ² βˆ’ 5)Β² = 25. Expand and solve β€” this becomes a quartic, and you'll need factoring.

Now consider xΒ² + y = 7 and xΒ² βˆ’ y = 3. Elimination is beautiful here. Adding directly: 2xΒ² = 10, xΒ² = 5, x = ±√5. Back-substitute: 5 + y = 7, y = 2. Two solutions: (√5, 2) and (βˆ’βˆš5, 2).

The catch: elimination can lose solutions when the eliminated variable was squared with a sign you discarded. If you square an equation to prepare it, extraneous roots can appear. Check every candidate solution in the original system. In the example above, both candidates check out, but skipping the check is how students lose points.

For nonlinear systems, always re-verify by plugging the numeric pair into both original equations. The algebra that produced the candidates may have introduced signs that aren't valid.

Setting It Up on Paper: A 90-Second Worked Check

A reliable layout saves time and catches errors. Here's a template that works on scratch paper and in exams.

Write the system at the top:

4x + 5y = 26 3x βˆ’ 2y = 1

Draw a vertical line to separate "work" from "check." On the left, write each multiplication:

Multiply eq1 by 2: 8x + 10y = 52 Multiply eq2 by 5: 15x βˆ’ 10y = 5 Add: 23x = 57, x = 57/23

Stop and look. 57 Γ· 23 = 2.478…, an ugly fraction. That's a warning sign. Either the problem is designed for a fraction answer, or the original coefficients don't cooperate well. In this case, multiply-check: does (57/23) satisfy both? Plug into eq2: 3(57/23) βˆ’ 2y = 1 β†’ 171/23 βˆ’ 2y = 1 β†’ 2y = 148/23 β†’ y = 74/23. Ugly, but exact. The solution is (57/23, 74/23).

If a problem gives back ugly fractions, the original coefficients might be designed so substitution is cleaner, or you may have misread a coefficient. Re-read the problem first. If it's still ugly, keep the fractions exact. Never round mid-solve.

For systems with decimals like 0.4x + 0.6y = 2 and 1.5x βˆ’ 0.5y = 3, scale both equations by 10 to clear decimals before eliminating. Multiply the first by 10 to get 4x + 6y = 20; multiply the second by 10 to get 15x βˆ’ 5y = 30. Now coefficients are integers and elimination proceeds normally.

Common Mistakes That Break Elimination

The errors below account for most lost points in algebra classes. Each is preventable with a check.

Signs during subtraction. When you subtract one equation from another, every term in the subtracted equation flips sign. Sliding a minus through 3x βˆ’ 4y = 7 becomes βˆ’3x + 4y = βˆ’7. Students frequently flip only the first term. Write the signs explicitly, then flip each one as a separate step.

Forgetting to scale every term. If you multiply 2x + 3y = 12 by 3, you must multiply 12 by 3 as well. The whole equation scales, not just the variable terms. Check the constant matches the scaling factor.

Eliminating the wrong pair. In a three-equation system, if you eliminate z using E1 and E2 first, then eliminate z using E2 and E3, you get F1 and F2 but they may share E2 in a way that also cancels x or y unintentionally. Deliberately pick E1 with E2 and E1 with E3 to keep symmetry.

Stopping at one variable. Getting x = 4 does not finish the problem. The solution is an ordered pair. Always back-substitute for the other variable and verify.

Rounding fractions early. Keep exact fractions throughout. 2.667 doesn't satisfy the equation the way 8/3 does, and small rounding errors compound.

Practicing on Real Problems: A Progression

Start with systems whose coefficients are already matched. Try x + y = 5 and x βˆ’ y = 1. Add: 2x = 6, x = 3, y = 2. Trivial. Move to systems needing one scaling: 2x + 3y = 12 and 4x βˆ’ 3y = 6 (worked above).

Then systems needing both equations scaled: 3x + 2y = 12 and 4x + 5y = 22. Multiply the first by 5 and the second by 2: 15x + 10y = 60 and 8x + 10y = 44. Subtract: 7x = 16, x = 16/7. Back-substitute into the first: 3(16/7) + 2y = 12 β†’ 48/7 + 2y = 12 β†’ 2y = 84/7 βˆ’ 48/7 = 36/7 β†’ y = 18/7. Check in the second: 4(16/7) + 5(18/7) = 64/7 + 90/7 = 154/7 = 22 βœ“.

Then three-variable systems. Follow the pattern from that section. Then special cases: parallel lines, dependent lines, systems with decimals, and systems where the solution includes negative numbers or zero. Zero is a valid solution; don't discard it.

Timing yourself is useful. A clean two-variable system takes about 45 seconds on paper once the routine is automatic. Three-variable systems take 3–5 minutes.

Solving by Elimination on a Calculator or Spreadsheet

Graphing calculators and spreadsheets can solve systems, but they don't teach the method. Use them for verification, not replacement.

On a TI-84 Plus or similar, enter the coefficient matrix in Matrix mode, then use the RREF function under Math β†’ Matrix to reduce it. The reduced matrix gives the solution directly if it exists. RREF stands for "reduced row echelon form," and its algorithm is essentially elimination performed by machine.

In Excel or Google Sheets, the function MDETERM computes the determinant of a coefficient matrix. If the determinant is zero, the system has no unique solution. Use MINVERSE and MMULT for full solutions: if A is the coefficient matrix and B the constants matrix, the solution is MINVERSE(A) Γ— B, entered as an array formula. A determinant of zero signals parallel or dependent lines, which requires the deeper check described earlier.

A failure to know: calculators and spreadsheets can't represent exact fractions well. The exact answer 16/7 becomes 2.285714…, and back-checking against the original equations produces small residuals that look like errors. For exact answers, work by hand or use a computer algebra system like SymPy in Python, which keeps fractions exact.

FAQ

Can elimination be used when the coefficients are fractions?

Yes, but clear the fractions first. Multiply each equation by the least common denominator of its coefficients to turn every fraction into an integer. For example, Β½x + β…“y = 4 becomes 3x + 2y = 24 after multiplying by 6. This step removes the main source of arithmetic errors and makes elimination practical.

What does the answer 0 = 7 mean after eliminating?

It signals an inconsistent system with no solution. The two equations describe parallel lines with the same slope and different y-intercepts. The contradiction 0 = 7 can never hold under any (x, y), so no point satisfies both equations simultaneously. Report the answer as "no solution."

Does elimination work for systems with three or more variables?

Elimination scales to any number of variables, but the number of steps grows. Four variables require using the first equation to eliminate a variable from the other three, then using the reduced system to eliminate another variable from its remaining two, and so on. Each variable removed reduces the system by one dimension. Expect 6–10 minutes of work for four-variable systems.

Which is better, elimination or substitution, for two-variable systems?

Elimination is faster when coefficients are close to matched or easily scaled. Substitution is faster when one variable has a coefficient of 1 or βˆ’1 or is already isolated. Neither is universally better. Inspect the system before choosing, and remember both methods give the same answer when applied correctly.

How do I check my elimination answer quickly?

Plug the solution into both original equations. If both produce true statements, the solution is correct. If one fails, the error is in the multiplication or subtraction step. Re-do the scaling arithmetic first, then the back-substitution. A quick check takes less than a minute and catches nearly every careless error.

Can elimination lose or create solutions in nonlinear systems?

It can create extraneous solutions if you square an equation during the process. It cannot lose solutions unless you divide by a variable expression that could be zero. The safeguard is to test every candidate solution in the original equations. Any candidate that fails is extraneous and should be discarded.

When does a system have infinitely many solutions?

When the two equations describe the same line, the elimination produces 0 = 0. Any (x, y) satisfying the surviving equation is a valid solution, so the answer is written as a parameterized pair β€” for instance, (x, 4 βˆ’ x) for any real x. Infinitely many solutions is a genuine answer, not a failure to solve.