The range of a relation is the set of all second coordinates (y-values) that actually appear among its ordered pairs. To find it, list every output value once, then remove duplicates and sort the result. The one condition that trips most students: a relation's range is defined only by the pairs it contains β there is no formula to evaluate and no domain rule to inherit.
That sounds simple until a relation is given as a rule like y = 1/(x β 2). Then "list the second coordinates" stops being a listing exercise and becomes an algebra problem about which outputs are reachable. Below are the four standard methods, worked in the format you'll meet them in class, plus the mistakes that cost marks on tests and the edge cases that break the shortcut answers.
The Definition: Where "Range" Actually Points in a Relation
A relation is any set of ordered pairs, and the range is the set of second elements across those pairs. For the relation {(1, 4), (2, 4), (3, 9)}, the range is {4, 9} β not {4, 4, 9}. Repeated outputs collapse to one entry because a set lists each value once.
Contrast that with the domain, which collects the first elements. In the same relation, the domain is {1, 2, 3}. Textbooks such as OpenStax's College Algebra treat both as sets, so braces and sorted order matter for full credit on written work.
Watch for a relation written as a table or a mapping diagram. A table with columns x and y behaves identically to a list of pairs: read down the y-column, not across the rows. A mapping diagram with arrows from inputs to outputs behaves the same way β collect every arrowhead's landing value.
One ambiguity deserves naming up front. "Relation" sometimes means a full equation with infinitely many pairs, and sometimes a finite set printed in braces. The listing methods below apply to the finite set; the algebraic methods apply to the equation. Decide which you have before you start, because the two demand different techniques.
How to Find the Range of a Relation Given as Ordered Pairs
Listing the second coordinates is the whole method, but presentation decides whether it's graded correct. Follow this order when a problem hands you pairs in braces.
1: Scan every ordered pair and copy only the second number onto scratch paper, in the order you meet it. 2: Cross out repeats, keeping one copy of each value. 3: Sort the survivors from least to greatest. 4: Write the answer in braces, commas between values: {β1, 0, 3, 8}. 5: Check that the count matches your scan β five pairs with three distinct outputs should yield three entries.
Take {(2, β1), (5, 0), (2, 3), (7, 3), (8, 8)}. The y-values in scanned order are β1, 0, 3, 3, 8. Removing the duplicate 3 and sorting gives {β1, 0, 3, 8}, four entries from five pairs.
The failure mode is copying the x-values by habit, especially when lines are long. If your answer's numbers all sit in the first position of their pairs, you found the domain. Re-scan pair by pair, reading right-hand numbers only, rather than reading down a list of pairs.
Testing favors this format because it's fast to mark. A multiple-choice question on a relation like the one above will usually include {2, 5, 7, 8} as a distractor β the domain dressed up as the range. The distractor is right there because so many students pick it.
Reading the Range Off a Graph: Project, Don't Count Points
On a graph, the range is every vertical position the relation touches, which you find by flattening the picture onto the y-axis. OpenStax's algebra sections describe this as projecting the graph onto the axis: imagine light shining sideways and read the shadow on the y-axis.
For a set of plotted points, the shadow is just the set of heights of the dots. Points at (1, 3), (1, 5), and (4, 3) give a range of {3, 5} β two values, because two dots share the height 3.
For a curve, the shadow is an interval or a union of intervals. A parabola opening upward with vertex at (1, β4) has range [β4, β). The square bracket means β4 is included; the parenthesis on β is standard convention, since infinity is a direction, not a number. A sideways parabola is the reverse case β its range runs to both infinities while its domain is the bounded one.
Arrows on the ends of a curve signal that it continues, so the interval should extend to infinity on that side. Don't trim it to fit the visible screen. A gap in the graph β a curve that skips a stretch of heights β splits the range into separate intervals joined by the union symbol βͺ.
Learn to read the extremes straight off the picture. The lowest point's height is the lower end of the range when the curve is continuous through it; the highest point's height is the upper end. Everything between is automatically included for an unbroken curve, so the only real work is locating those two heights.
An Equation's Range: Two Worked Examples That Expose the Excluded Value
With an equation, the range is the set of outputs the rule can produce. For y = 1/(x β 2), running inputs inward toward x = 2 sends the output outward without limit: x = 3 gives 1, x = 2.1 gives 10, x = 2.01 gives 100. Running from the other side, x = 1 gives β1 and x = 1.9 gives β10, so negatives appear too β but never 0.
Solving for x makes the exclusion explicit. From y = 1/(x β 2), cross-multiplying gives y(x β 2) = 1, then x = 1/y + 2. That expression is undefined at y = 0, so 0 cannot be an output. The range is (ββ, 0) βͺ (0, β).
A quadratic behaves differently. For y = xΒ² β 6x + 11, the vertex form is (x β 3)Β² + 2, so the vertex sits at (3, 2) and the parabola opens upward. The minimum output is 2, achieved at x = 3, and outputs climb from there without limit: range [2, β).
Negatives never appear here, which is the reverse of a common expectation. Students who graph quickly and see the parabola dip below the x-axis often assume a negative minimum, but the y-intercept of 11 and vertex height of 2 both sit above zero β the curve crosses the axis nowhere.
The pattern to internalize: denominators throw out outputs that would force a division by zero, and even roots throw out negative outputs beneath them. Square roots give ranges starting at 0 once the radicand is nonnegative, and rational functions typically give the whole number line minus one or two holes. Verify the hole by solving for the input and seeing where that expression fails.
The Fast Method, the Failure Case, and the 94% Trap
For a quadratic in standard form, the vertex's y-coordinate is your range endpoint. For y = axΒ² + bx + c, compute h = βb/(2a) and then the minimum or maximum value is c β bΒ²/(4a). For y = xΒ² β 6x + 11, that's 11 β 36/4 = 2, matching the vertex-form answer above. One computation, one endpoint, done.
That shortcut is also where the single most common failure case lives: using it inside a restricted domain. If a company's profit model is P = βxΒ² + 40x β 300 for 0 β€ x β€ 10 units, the vertex at x = 20 lies outside the allowed inputs. The maximum profit within the stated window comes from testing the endpoints: x = 0 gives β300 and x = 10 gives 0, so the range is [β300, 0] β not the vertex value of 100 that the unrestricted parabola would suggest.
Now the trap. A well-known classroom demonstration treats a relation of 50 ordered pairs in which 3 of the second coordinates are missing, so 47 pairs have real outputs β and asks what the range is. The correct response is that a missing output is not an output: the range still lists only the 47 present values, and the "94% present" figure describes the data collection, not the range. The three missing pairs contribute nothing to either set, and including placeholder values like 0 for them is the classic error.
Standardized-test scoring rewards the literal reading. Missing data stays missing; it never becomes a member of the set. If a question describes a relation with some pairs unfilled, list only the values actually shown.
Domain, Range, and Codomain: The Comparison Rows That Settle the Confusion
Three terms get tangled constantly, and a single table untangles them faster than paragraphs. For the relation y = xΒ² with real inputs, here's how each set looks.
| Set | Values for y = xΒ² | Interval | Determined by |
|---|---|---|---|
| Domain | every real input | (ββ, β) | What you're allowed to put in |
| Range | 0 and every positive real | [0, β) | What actually comes out |
| Codomain | every real number | (ββ, β) | The declared target set |
The codomain is the container, the range is what gets filled inside it. A function can be onto when its range equals its codomain, and that's a property of the function, not a routine calculation for a relation.
The direction of the bracket matters on the same row. A closed bracket includes the endpoint; a parenthesis excludes it. A hole at y = 3, coming from a denominator like (y β 3), reads as (ββ, 3) βͺ (3, β). An output the rule reaches exactly once, like the endpoint of a square root's range at 0, gets a closed bracket: [0, β).
One admitted conflict: notation for the domain can be written as an inequality ("x β₯ 0"), as an interval, or in set-builder form. Schools differ, so confirm which the teacher expects. The values are identical; only the costume changes.
Get the membership right and the terms sort themselves. The range is a subset of the codomain, always β even when the two happen to be equal. Writing a relation's range as its codomain is the mistake that looks sophisticated and is still wrong.
Special Relations: Roots, Reciprocals, and the Vertical Line Test
Square roots, reciprocals, and lines each carry a range rule worth memorizing, because re-deriving takes time on a test. The table below summarizes the four families you'll meet most.
| Relation | Range | Reason | Endpoint included? |
|---|---|---|---|
| y = β(x) | [0, β) | Principal square root is never negative | Yes, at 0 |
| y = β(xΒ² β 1) | [0, β) | Radicand reaches 0 at x = Β±1 | Yes |
| y = 1/x | (ββ, 0) βͺ (0, β) | Reciprocal never equals 0 | No, hole at 0 |
| y = x (a line) | (ββ, β) | Every height occurs | Not applicable |
The vertical line test is a domain test, not a range test β it asks whether a graph is a function by checking that no vertical line hits twice. The horizontal line test does for the range what the vertical test does for the domain. A function is one-to-one if no horizontal line crosses its graph twice, which corresponds to each output having a single input. That test doesn't change the range, but it tells you whether the relation passes a key property, like being invertible.
A line with slope zero needs care. The relation y = 5 is a function whose range is the single value {5}, not an interval. The domain is still every real input, and graphing yields a flat horizontal line at height five. Students who write [5, β) have misread the graph as a ray.
A vertical line, by contrast, isn't a function at all β the same input has infinitely many outputs. Its range is all real y-values, but it's the exception to every rule-of-thumb about slope. Recognize the exception so you don't apply a function rule to a non-function.
Check Your Answer: Four Tests That Catch Range Mistakes
Range errors are usually errors of omission rather than arithmetic. Four quick checks catch most of them before the paper leaves your desk.
1: Test each output range. For a domain you can't easily inspect, or an unfamiliar curve, test a few y-values by solving the equation for x. If a chosen y-value demands dividing by zero or taking a square root of a negative, that y-value falls outside the range. 2: Sketch the graph when any doubt remains. A quick plot shows the minimum, maximum, and any skipped stretches; a table of six points usually suffices. 3: Confirm the bracket type at each endpoint. Holes are parentheses, solid points are brackets, and arrows mean infinity. 4: Cross-check the domain for a hidden constraint you may have missed. A restricted domain narrows the range, so reapply the endpoint test when the domain isn't all reals.
A worked example ties these together. For y = (x + 1)/(x β 4), solving for x gives x = (β1 β 4y)/(y β 1). The denominator forces y β 1, and y = 1 is the horizontal asymptote the graph flattens toward without touching. Range: (ββ, 1) βͺ (1, β). The table of values would show the graph rising to 1 from below and falling toward 1 from above, never landing on it.
The endpoint values themselves are always in play as answers, even if a wrong one looks similar. An answer that says (1, β) instead of (ββ, 1) βͺ (1, β) has missed half the graph β the half the table only shows if you look at x < 4 as well as x > 4.
Common Range Mistakes and What They Cost
Four mistakes account for most lost marks on range questions. Each has a cheap fix.
Listing the domain: reading the first coordinate by habit. Fix: read the second entry in each pair deliberately. Treating infinity as a number: writing a closed bracket at β. Fix: infinity always takes a parenthesis. Assuming no repeats: writing every output even when duplicates occur. Fix: sets collapse duplicates; count distinct values against the pair count. Ignoring the codomain vs. range distinction: declaring the range of y = xΒ² to be all reals because the codomain is all reals. Fix: let the graph or vertex decide the range; the codomain is the container, never the answer.
The information-gain comparison across domestic and foreign school systems is worth a sentence, because notation habits differ. Some curricula in the US and UK write ranges as inequalities and rarely use union symbols; others, especially those following the International Baccalaureate, treat set notation and interval notation as interchangeable and expect some awareness of the difference. Where the teacher demands a notation, use it; where either is accepted, interval notation saves space.
One last member of the set has to be accounted for before the answer is complete: the endpoint bracket. Standardized questions frequently include both the open and closed version of the same endpoint in the choices, so the bracket is the whole test. Get it right only if the endpoint is actually a reachable output β not merely the value the graph approaches.
FAQ
Can a range include 0?
Yes, if 0 is actually produced by the relation β for y = βx, the range includes 0 because β0 = 0, while y = 1/x excludes it since no input makes 1/x equal 0. The bracket versus parenthesis at 0 is the entire distinction.
What is the range of a relation with repeated outputs?
Repeats collapse: {(1, 5), (2, 5), (3, 9)} has range {5, 9}, not {5, 5, 9}. If two ordered pairs have the same second coordinate, that value appears once in the range.
Why do some teacher cas write the range of y = xΒ² as [0, β) instead of y β₯ 0?
Both are correct, and schools differ on which form they expect. Interval notation is compact; inequality notation tends to be the required form in US high school algebra and some British curricula. Confirm your teacher's preference before an exam.
What happens to the range if the domain is restricted, like 0 β€ x β€ 10?
Restriction narrows the range, sometimes severely. A parabola whose vertex lies outside the input window needs endpoint testing rather than vertex reading: check both x = 0 and x = 10, take the smaller and larger outputs, and close the intervals at both ends.
Is the range the same as the codomain?
No β the codomain is the declared container, and the range is what actually comes out. For y = xΒ², the codomain might be all reals, yet the range is [0, β); every range is a subset of its codomain, never the reverse.
How do you find the range of a relation given as a graph?
Project the graph onto the y-axis: the minimum and maximum heights bracket the answer, and any vertical gap splits the result into separate intervals. Unlike a finite list of pairs, a continuous graph usually yields an interval or a union of intervals rather than a set of scattered numbers.
What does a missing output do to the range β does it stay as a gap?
A relation missing an output simply doesn't include that value β it's not a gap to fill in, and placeholder zeros are wrong. The range lists only values that are actually present; the 94% figure in the common classroom example describes data completeness, not the range itself.