The range of a function is the set of all possible output values it can produce, and you find it by determining every y-value the rule can actually reach for inputs in its domain. For most functions you'll meet in algebra and calculus, that means combining the domain, the function's behavior at its edges, and its turning points. The method below covers polynomials, rational functions, radicals, exponentials, and trigonometry, with the specific failure cases that trip people up.

Start by Naming the Domain, Because Range Lives Inside It

You can't state a range until you know which inputs are legal, since every excluded x-value can remove an output from the range or open a gap in it. Take f(x) = 1/x. The domain excludes x = 0, and the range consequently excludes y = 0, because no finite input makes the reciprocal equal zero. That single excluded value is the whole difference between the correct answer and a careless "all real numbers."

Two domain restrictions do most of the work in textbook problems. Division by zero removes the x-values that zero a denominator, and even roots of negative numbers remove the x-values that make the radicand negative. A third, less visible restriction comes from the applied context: if x counts physical objects, negative and fractional inputs may be meaningless even though the algebra accepts them.

Write the domain first, in interval notation, before you touch the range. For f(x) = sqrt(x βˆ’ 3), the domain is [3, ∞), and that lower bound of 3 immediately tells you the smallest possible output is f(3) = 0. Skipping this step is the most common reason students report a range that includes values the function can never produce.

Applied problems tighten this further. A function modeling the area of a rectangle with 20 meters of fencing, A(x) = x(10 βˆ’ x), has a mathematical domain of all reals but a physical one of (0, 10). According to the standard modeling convention taught in most algebra curricula, the context-defined domain governs, so the range must be computed on (0, 10) rather than on the full real line.

The Vertex Formula Gives You the Range of Any Quadratic

For a quadratic in standard form f(x) = axΒ² + bx + c, the vertex sits at x = βˆ’b/(2a), and its y-coordinate is the highest or lowest output the function ever produces. Compute that y-value and you have one endpoint of the range; the parabola's direction supplies the rest. This single formula handles every quadratic you'll see, including ones where factoring fails.

Work through f(x) = 2xΒ² βˆ’ 8x + 5. Here a = 2, b = βˆ’8, so the vertex x-coordinate is βˆ’(βˆ’8)/(2Β·2) = 2. Substituting gives f(2) = 8 βˆ’ 16 + 5 = βˆ’3. Because a = 2 is positive, the parabola opens upward, so the range is [βˆ’3, ∞). You can verify with two symmetric points: f(1) = βˆ’1 and f(3) = βˆ’1, both above βˆ’3, confirming the vertex is a minimum.

When a is negative, the same computation produces a maximum and the range becomes (βˆ’βˆž, k], where k is the vertex y-value. For f(x) = βˆ’xΒ² + 4x, the vertex is at x = 2 with f(2) = 4, so the range is (βˆ’βˆž, 4]. The sign of a alone decides whether the bracket goes on the left or the right.

Watch for quadratics presented in vertex form, f(x) = a(x βˆ’ h)Β² + k. The range is [k, ∞) when a is positive and (βˆ’βˆž, k] when a is negative, no computation required beyond reading k. That shortcut is worth memorizing because vertex form appears constantly in optimization problems.

A failure case worth flagging: if you're given a restricted domain, the vertex may lie outside it, and then the vertex value is not in the range at all. For f(x) = xΒ² on [2, 5], the vertex at x = 0 is irrelevant, and the range is [4, 25], computed from the endpoints of the restricted interval.

Check the End Behavior Before You Trust an Endpoint

End behavior tells you whether a function's outputs stretch to infinity, and it's the reason polynomial ranges almost always run to one or both infinities. For any polynomial of odd degree, the two tails head in opposite directions, so the range is all real numbers. For even degree, both tails head the same way, so the range is bounded on one side only.

Consider f(x) = xΒ³ βˆ’ 4x. As x β†’ ∞, f(x) β†’ ∞, and as x β†’ βˆ’βˆž, f(x) β†’ βˆ’βˆž. The cubic is continuous, so by the intermediate value theorem it hits every real number between its tails. The range is (βˆ’βˆž, ∞), and no amount of local wiggle at the turning points changes that, because the tails escape in both directions.

Even-degree polynomials behave differently. For f(x) = x⁴ βˆ’ 2xΒ², both tails go to +∞, so the range has a lower bound but no upper bound. Finding that lower bound requires the derivative or a substitution; setting u = xΒ² turns the function into uΒ² βˆ’ 2u, a quadratic with minimum βˆ’1 at u = 1, giving a range of [βˆ’1, ∞). The end behavior told you the shape of the answer before you computed the number.

Rational functions need their own end-behavior check, because horizontal asymptotes cap the tails. For f(x) = (2x + 1)/(x βˆ’ 3), the degree of numerator and denominator match, so the horizontal asymptote is y = 2, the ratio of leading coefficients. The function never equals 2, which removes that value from the range even though the tails approach it.

That asymptote rule is where most range errors in precalculus hide. A student who graphs the function sees the curve flattening toward 2 and writes the range as (βˆ’βˆž, ∞). The correct answer excludes 2 explicitly, and the exclusion comes from solving the equation y = (2x + 1)/(x βˆ’ 3) for x and discovering that y = 2 produces no solution.

Solve for x to Read the Range Directly Off the Inverse

Setting y = f(x) and solving for x turns the range question into a domain question on the inverse relation, and it catches excluded values that graphs make easy to miss. If the algebra produces a valid x for a given y, that y is in the range; if it produces a contradiction or a division by zero, it isn't. This approach is the most reliable general method for rational and algebraic functions.

Try f(x) = (x + 3)/(x βˆ’ 5). Set y = (x + 3)/(x βˆ’ 5), multiply through to get y(x βˆ’ 5) = x + 3, then collect terms: x(y βˆ’ 1) = 5y + 3, so x = (5y + 3)/(y βˆ’ 1). That expression is defined for every y except y = 1, so the range is (βˆ’βˆž, 1) βˆͺ (1, ∞). The excluded value falls out of the algebra automatically.

The method also exposes cases where the range is narrower than the domain suggests. For f(x) = xΒ²/(xΒ² + 1), solving y(xΒ² + 1) = xΒ² gives xΒ² = y/(1 βˆ’ y), which requires 0 ≀ y < 1 for a real x to exist. The range is [0, 1), a bounded interval that a quick sketch might render as "all positive numbers."

Failures show up as impossible equations. If solving for x yields something like xΒ² = βˆ’4 for a particular y, that y is outside the range. If it yields x = 1/(y βˆ’ 3), then y = 3 is excluded. Read the resulting formula's own domain restrictions and you have the range.

One caution applies to non-injective functions. Squaring, for instance, loses information, so a quadratic will produce two branches when you solve for x. Both branches must be checked against the domain; a branch that produces x-values outside the domain doesn't certify the corresponding y-value. For f(x) = xΒ² on [0, ∞), solving gives x = √y, and the requirement x β‰₯ 0 forces y β‰₯ 0, which matches the expected range.

Radical Functions Inherit Their Lower Bound From the Radicand

A square root's output can never be negative, so every radical function has a floor at zero, shifted by whatever constant sits outside the root. For f(x) = sqrt(x βˆ’ 3) + 2, the radicand requires x β‰₯ 3, and at x = 3 the output is 2. As x grows, the root grows without bound, so the range is [2, ∞). The +2 outside the radical shifts the floor from 0 to 2.

Cube roots behave differently because they accept negative inputs and produce negative outputs. For f(x) = βˆ›(x + 1), the domain is all reals, and the range is all reals, since the cube root sweeps every real value as its argument does. Students who apply the square-root floor to a cube root get a range that's too narrow by half.

Even roots with coefficients need one extra step. For f(x) = βˆ’2√(x + 4), the negative coefficient flips the direction: at x = βˆ’4 the output is 0, and as x increases the output decreases without bound, so the range is (βˆ’βˆž, 0]. The coefficient's sign, not the root's, decides whether the interval runs up or down.

Here's a failure case that appears on exams regularly. For f(x) = √(4 βˆ’ xΒ²), the domain is [βˆ’2, 2], and the radicand peaks at 4 when x = 0. The range is therefore [0, 2], not [0, ∞), because the input itself is bounded. Bounding the input always bounds a monotone function's output, and the endpoints of the domain are where you check.

Exponential and Logarithmic Functions Break the All-Reals Habit

Exponentials such as f(x) = 2^x have a horizontal asymptote at y = 0 and rise without bound, so the range is (0, ∞). Every exponential with a positive base follows this pattern, and transformations shift the asymptote rather than eliminating it. For f(x) = 3Β·2^x βˆ’ 5, the asymptote moves to y = βˆ’5 and the range becomes (βˆ’5, ∞).

Logarithms invert the picture. For f(x) = log(x βˆ’ 1), the domain is x > 1 and the range is all real numbers, because a logarithm reaches every real output as its argument sweeps the positive reals. That's the mirror image of the exponential result, and it's the cleanest illustration of why domain and range swap under inversion.

Reflections change the interval's direction without changing its openness. For f(x) = βˆ’e^x + 4, the negative coefficient reflects the exponential downward, so the asymptote at y = 4 is an upper bound and the range is (βˆ’βˆž, 4). The asymptote never gets touched, which is why the interval stays open at 4.

These two families are worth pairing because they catch a common error: assuming that a function with an asymptote must have a range excluding infinitely many values. An exponential excludes one boundary and reaches everything on one side of it; a logarithm excludes nothing at all. The asymptote's location tells you where the boundary sits, and the function's monotonicity tells you which side of it the outputs live.

Trigonometry Needs Amplitude and Vertical Shift, Not Endpoints

Sine and cosine oscillate between fixed bounds, so their ranges are closed intervals determined by amplitude and vertical shift. For f(x) = 3 sin(x) + 2, the sine term runs from βˆ’1 to 1, multiplying by 3 gives βˆ’3 to 3, and adding 2 gives βˆ’1 to 5. The range is [βˆ’1, 5], and both brackets are closed because the extremes are attained.

Tangent is a different animal. For f(x) = tan(x), the range is all real numbers, because the tangent sweeps every real value between consecutive vertical asymptotes, and there are infinitely many such intervals in the domain. Students often assume tangent is bounded like sine; it isn't, even though its graph looks confined on a calculator screen.

Phase shifts and horizontal stretches don't affect the range. For f(x) = 4 cos(2x βˆ’ Ο€) βˆ’ 1, the amplitude is 4 and the vertical shift is βˆ’1, so the range is [βˆ’5, 3] regardless of the phase. Only the coefficient in front and the constant at the end matter for the output interval.

A worked exam-style case: f(x) = 2 sin(3x) + 1 on the restricted domain [0, Ο€/6]. Over that interval the argument 3x runs from 0 to Ο€/2, and sine rises from 0 to 1 monotonically. The output therefore runs from 2Β·0 + 1 = 1 to 2Β·1 + 1 = 3, giving a range of [1, 3]. Note that the full-function range of [βˆ’1, 3] would be wrong here because the restricted input never lets the sine dip below zero.

A Comparison Table for the Function Families You'll Meet

The same seven-step routine applies to every family, but the decisive feature differs, and knowing which feature to check saves time on exams. Use this table as a quick reference after you've computed the domain.

Family Example Decisive feature Range
Linear f(x) = 3x βˆ’ 2 Nonzero slope, domain all reals (βˆ’βˆž, ∞)
Quadratic f(x) = xΒ² βˆ’ 6x + 5 Vertex at (3, βˆ’4), opens upward [βˆ’4, ∞)
Cubic f(x) = xΒ³ βˆ’ x Odd degree, tails escape both ways (βˆ’βˆž, ∞)
Rational f(x) = (x + 1)/(x βˆ’ 2) Horizontal asymptote y = 1 excluded (βˆ’βˆž, 1) βˆͺ (1, ∞)
Square root f(x) = √(x βˆ’ 4) + 1 Radicand floor at 0, shift +1 [1, ∞)
Cube root f(x) = βˆ›(x βˆ’ 2) Odd root accepts all inputs (βˆ’βˆž, ∞)
Exponential f(x) = 2^x + 3 Asymptote at y = 3, rises (3, ∞)
Logarithm f(x) = log(x + 5) Inverts exponential range (βˆ’βˆž, ∞)
Sine f(x) = 4 sin(x) βˆ’ 2 Amplitude 4, vertical shift βˆ’2 [βˆ’6, 2]
Tangent f(x) = tan(2x) Full sweep between asymptotes (βˆ’βˆž, ∞)

The linear row contains an edge case worth naming. A horizontal line f(x) = 5 has a range of just {5}, a single value, even though its domain is all reals. The table's "nonzero slope" qualifier is doing real work there, and a constant function is the one case where a "linear" function fails to produce all reals.

Rational functions with a numerator degree exactly one higher than the denominator have no horizontal asymptote but do have a slant asymptote, and their range is typically all reals. For f(x) = (xΒ² + 1)/x, the slant asymptote is y = x, and the function reaches every real y, though the algebraic check takes a few more lines. Confirm such cases with the solve-for-x method rather than with a sketch.

Five Mistakes That Produce the Wrong Range and How to Catch Them

The errors below account for most lost points on range questions, and each one has a quick self-check. Catching them before you write the final interval is faster than re-deriving the answer.

1: Reporting all reals because the graph "looks continuous." Check for horizontal asymptotes and for excluded values produced by solving for x. If either exists, the range is not all reals. 2: Using the domain as the range. Domain restrictions constrain inputs; only the function's values constrain outputs. Confirm by computing f at the domain's endpoints. 3: Forgetting that an asymptote is excluded. Write the range with parentheses at any y-value produced by a horizontal asymptote, unless you can show the function actually attains it. 4: Applying square-root floors to cube roots or other odd roots. Odd roots reach every real value; the floor at zero is specific to even roots. 5: Ignoring a restricted domain. When the problem gives an interval for x, recompute both endpoints of the output and check whether the vertex or maximum lies inside the interval before using it.

The asymptote mistake deserves a second look because it hides in plain sight. For f(x) = x/(xΒ² + 1), solving for x gives a quadratic whose discriminant must be non-negative, which restricts y to [βˆ’1/2, 1/2]. A student who notices that the function's values look small on a graph might guess "between βˆ’1 and 1," which is close but wrong by a factor of two on each end.

A time computation makes the value of the checks concrete. If a typical exam gives you 90 minutes for 25 problems, about 3.6 minutes per problem, a solve-for-x verification on a rational function takes roughly 45 seconds while re-graphing the function and re-reading the range off a sketch takes over 2 minutes. The algebraic check is the cheaper habit, and it's the one that catches asymptote exclusions reliably.

FAQ

Can a function have an empty range?

No, a function from a nonempty domain always produces at least one output, so its range contains at least one value. The only way to get an empty range is to define the function on an empty domain, which is a technical edge case rather than a typical exam scenario. For any function you'll be asked about, check the domain first and you'll never report an empty range.

Is the range of a cubic function always all real numbers?

Every cubic polynomial has range (βˆ’βˆž, ∞), because odd-degree polynomials have tails escaping in opposite directions and are continuous, so they cross every real y-value at least once. The number of turning points doesn't matter, since they only create local wiggles. The intermediate value theorem guarantees the full sweep.

How do I find the range of a piecewise function?

Compute the range of each piece on its own subdomain, then take the union of the resulting intervals. For f(x) = xΒ² on [0, 2] and 3x on (2, 5], the first piece gives [0, 4] and the second gives (6, 15], so the range is [0, 4] βˆͺ (6, 15]. Check whether the pieces' ranges overlap or leave gaps before writing the union.

What's the difference between range and codomain?

The codomain is the set the function is declared to map into, while the range is the subset of the codomain that's actually hit. For f: ℝ β†’ ℝ defined by f(x) = xΒ², both the domain and codomain are all reals, but the range is only [0, ∞). Textbooks sometimes call the codomain the "target set," and it's usually chosen by convention rather than computed.

Why does the range of 1/x exclude zero?

Setting y = 1/x and solving gives x = 1/y, which is undefined at y = 0. No finite real input makes the reciprocal zero, since a fraction equals zero only when its numerator is zero, and here the numerator is the constant 1. The exclusion is structural, not a graphing artifact, and it survives every transformation of the form a/(x βˆ’ h) + k, which excludes only y = k.

Can a quadratic's range include its vertex value if the domain is restricted?

Only if the vertex's x-coordinate lies inside the restricted domain. For f(x) = xΒ² βˆ’ 4x + 1 on [1, 5], the vertex is at x = 2 with value βˆ’3, which is inside the interval, so the range is [βˆ’3, 12], with 12 coming from f(5). Shift the interval to [3, 5] and the vertex falls outside, making the range [βˆ’2, 6] with the minimum at x = 3.