An improper integral in AP Calculus BC is any definite integral in which one of the standard preconditions of the Fundamental Theorem of Calculus fails. Either an integration bound is infinite, or the integrand becomes unbounded somewhere on the interval, or both happen at once. The course defines two clean families, often labelled Type 1 and Type 2, and BC candidates are expected to evaluate them by replacing the offending bound or point with a finite limit, evaluating the resulting proper integral, and then taking the limit. On the Free Response section, the work is graded row by row: the limit setup, the antiderivative row, the evaluation row, and the final numerical answer each carry their own scoring decision. This article walks through how those rows actually read on an AP Calculus BC exam, where students typically lose points, and how to plan the calculation so that the limit behaviour is visible to the reader before the numerical answer is even computed.
The two shapes AP Calculus BC tests: infinite bounds versus unbounded integrands
Every improper integral on the AP Calculus BC exam reduces to one of two shapes, and the first scoring decision the rubric makes is whether the student identified the correct shape. Type 1 improper integrals involve an integration bound of infinity or negative infinity while the integrand stays finite and continuous on the rest of the interval. The classic example is the integral from 1 to infinity of 1 over x squared, which converges to 1. The setup row requires the student to rewrite the expression as a limit, typically lim as b approaches infinity of the integral from 1 to b of the integrand, and the scoring reader is watching for two things: a limit sign on the page, and the same integrand written inside a finite integral with a letter-based upper bound. A bare antiderivative evaluation across an infinite range, without the limit wrapper, does not earn the first rubric row in most years of the exam.
Type 2 improper integrals involve a finite interval where the integrand has a vertical asymptote, a removable discontinuity that becomes non-removable, or some other unbounded behaviour. The integral from 0 to 1 of 1 over the square root of x is the textbook instance. The setup row here is a sum of one-sided limits, expressed as lim as a approaches 0 from the right of the integral from a to 1 plus, in some problems, a symmetric term for a discontinuity at the interior point. If the discontinuity lives at an interior point c, the student is expected to split the integral at c and write two limits, one on each side, before evaluating. Skipping the split and writing a single symmetric limit silently assumes the integrand is well behaved, which on a BC exam is precisely the assumption that is being tested.
Students preparing for the FRQ should keep a mental note of the third, rarer shape in which both an infinite bound and an interior discontinuity appear in the same integral. The integral from 0 to infinity of 1 over the square root of x plus 1, evaluated by writing lim as a approaches 0 from the right, then lim as b approaches infinity of the integral from a to b, is a two-stage problem. The rubric tends to award the setup row if both limits appear in the correct order, and the convergence claim only earns the final row if both inner limits resolve. The key tactical point: when the integrand is well behaved at the troublesome bound, candidates should write the limit anyway. Leaving the limit off is one of the most common ways to lose the first row of an AP Calculus BC improper integral problem.
How the setup row is read
On an AP Calculus BC FRQ, the reader does not award points for "understanding" in the abstract. The reader scans for symbolic evidence. For a Type 1 problem, the symbols that matter are the limit notation, the letter b or t used as a placeholder for the infinite bound, and the integrand written unchanged inside the new finite integral. For a Type 2 problem, the symbols that matter are the letter a (or any auxiliary letter) approaching the discontinuity, a one-sided direction written explicitly with a plus or minus sign, and the integrand written on a closed interval that does not contain the asymptote. If a student writes the antiderivative and substitutes a symbolic bound but never shows the limit wrapper, the reader can only award points for the antiderivative row, not the setup row, and that typically costs one rubric point on a six-point FRQ.
Evaluating Type 1: the limit-of-integration row on infinite bounds
Once a Type 1 improper integral has been rewritten as a limit, the next scoring row is the antiderivative. For convergent examples like the integral from 1 to infinity of 1 over x squared, the antiderivative is negative 1 over x, and the candidate is expected to evaluate it at the upper limit b and the lower limit 1, subtract in the correct order, and then take the limit as b approaches infinity. The limit step, not the antiderivative step, is where most of the work happens, and the reader is watching for a clean argument that the limit exists and is finite. On a BC exam, that argument can take one of three forms: a direct evaluation by substitution, an application of L'Hospital's rule to a 0/0 or infinity/infinity indeterminate form, or a p-integral comparison written as a sentence rather than a formula.
For a question like the integral from 1 to infinity of 1 over x to the p power, the candidate is expected to recognise the p-test, write the antiderivative as x to the 1 minus p over 1 minus p, evaluate at b, and then take the limit. The limit exists and is finite exactly when p is greater than 1. If p is less than 1 or p equals 1, the limit diverges to infinity or fails to exist, and the candidate is expected to state that the integral diverges rather than to invent a finite value. On a BC FRQ, divergence is a perfectly acceptable final answer, and the scoring reader awards the convergence row only when the candidate has correctly classified the integral as convergent or divergent, not when they have produced a numerical value. A student who writes "equals infinity" without the limit setup usually loses two rows: the setup row and the convergence claim row.
For more complex integrands, the limit step often becomes a L'Hospital calculation. Consider the integral from 1 to infinity of ln x over x cubed. The antiderivative can be obtained by integration by parts, yielding a fraction with x cubed in the denominator, and the limit as x approaches infinity of that fraction is an infinity-over-infinity form that resolves cleanly with one application of L'Hospital's rule. The scoring reader is watching for the candidate to state that the original form is indeterminate, apply L'Hospital at least once, and produce a finite limit. Skipping the indeterminate-form justification and writing only the final numerical value is a common way to lose the L'Hospital row. On AP Calculus BC FRQs, the rubric for L'Hospital's rule typically requires a visible statement that the limit is indeterminate in a recognised form, and a candidate who omits that statement cannot earn the row even if the algebra is correct.
Tactical sequencing for Type 1 problems
In my experience marking through past BC exam solutions, the strongest candidates write the limit-of-integration line in three steps rather than two. They write the limit, they write the antiderivative evaluated at the symbolic bound, and they write the simplified expression whose limit is being taken. Only after those three lines appear does the convergence argument enter the page. This sequencing matters because the reader is scoring symbolically: the limit sign, the antiderivative, the evaluation at the symbolic bound, and the simplified limit expression are each separate rows on the rubric, and a student who compresses them into one long equation is at risk of having the reader miss a row entirely. For a 15-minute FRQ on improper integrals, the three-line setup is roughly two minutes of work, and it is the cheapest two minutes on the exam in terms of points per minute spent.
Evaluating Type 2: unbounded integrands and the split-integral decision
Type 2 improper integrals look superficially similar to Type 1 because both involve a limit, but the limit is on a bound approaching a finite point rather than infinity. The integral from 0 to 1 of 1 over the square root of x, for instance, requires the candidate to write lim as a approaches 0 from the right of the integral from a to 1 of x to the negative one half. The integrand is unbounded near 0, but the antiderivative 2 times the square root of x is bounded as a approaches 0 from the right, and the limit is 2. The crucial scoring row is the one-sided direction on the limit: writing a plain limit, without the plus sign, is treated by the rubric as if the candidate had not identified the discontinuity at all. On a BC exam, that typically costs the setup row, and the antiderivative row is the only one that survives.
When the unbounded behaviour occurs at an interior point of the interval, the candidate must split the integral. Consider the integral from negative 1 to 1 of 1 over x to the two thirds power. The integrand is unbounded at x equals 0, which is interior to the interval, so the candidate is expected to write the original integral as the sum of two pieces, one from negative 1 to a and one from a to 1, with a approaching 0 from the left in the first piece and from the right in the second. Both inner limits must be taken, and both must be finite, for the integral to converge. If only one of the two inner limits is finite, the integral diverges, and the candidate is expected to say so. The split is a scoring decision in its own right: a candidate who writes a single symmetric limit without splitting is implicitly assuming that the function is integrable in a principal-value sense, which is not the convention on the AP Calculus BC exam.
For integrands with multiple discontinuities on the interval, the candidate must split at every discontinuity. The integral from 0 to 3 of 1 over x minus 1 has an asymptote at x equals 1, so the correct setup is the integral from 0 to 1 plus the integral from 1 to 3, each handled as a separate limit. Skipping the split and writing the antiderivative as the natural log of absolute value of x minus 1 evaluated from 0 to 3 is a frequent mistake, because the antiderivative happens to exist, but the integral does not exist as a single convergent quantity. The scoring reader can see the antiderivative and may award the antiderivative row, but the convergence row and the numerical row both fail because the integral diverges at x equals 1. This is one of the most expensive single errors on a BC FRQ, and students preparing for the exam should drill the split step on every Type 2 problem where the asymptote sits inside the interval.
Why L'Hospital is not the first move on Type 2 problems
Unlike Type 1 problems, Type 2 problems often resolve without L'Hospital's rule at all. The antiderivative of a simple unbounded integrand like x to the negative one half is well known, and the limit of the antiderivative at the asymptote can be taken by direct substitution. Pulling out L'Hospital on a Type 2 problem is usually a sign that the candidate has not identified the asymptote, has computed an antiderivative that does not exist, or has set up the problem incorrectly. On a BC exam, an unnecessary L'Hospital step on a Type 2 problem is a yellow flag to the reader that the candidate may be computing a derivative problem rather than an integral problem, and the rubric reserves the convergence row for problems where the limit is taken on an antiderivative expression, not on a derivative expression. If the candidate is not sure which technique to use, the safer move is to compute the antiderivative, evaluate at the symbolic bound, and take the limit by inspection before reaching for L'Hospital.
Convergence claims and the divergence language AP readers expect
The convergence row on an AP Calculus BC improper integral FRQ is one of the most stereotyped rows on the exam. The reader is looking for one of two short statements: "the integral converges to [number]" or "the integral diverges." Anything more elaborate is usually a waste of words, and anything less is usually insufficient. Phrases like "the limit goes to infinity" without the word "diverges" lose the row, because the rubric ties the row to the specific vocabulary. Candidates preparing for the exam should memorise the two acceptable formats and use them verbatim on the FRQ. In my experience, the convergence row is the row that students most often score correctly when they have a clear answer and most often miss when they are guessing, and the difference is almost always the verb choice in the final sentence.
For integrals that diverge, the candidate is expected to identify the type of divergence. Saying "the integral diverges to infinity" is a stronger statement than "the integral diverges," and on questions where the divergence type is part of the question, the stronger statement is required. For integrals that oscillate, like the integral from 0 to infinity of sin x over x, the candidate is expected to say that the limit does not exist, not that the integral diverges to infinity. On a BC exam, oscillation is treated as a separate category from divergence to infinity, and the rubric will not award the convergence row to a student who conflates the two. The distinction matters because the integrals that oscillate often have a different antiderivative structure, and the candidate who recognises the oscillation can use the Dirichlet test or a comparison argument to score the convergence row even when the antiderivative does not exist in closed form.
For convergent integrals, the numerical value must be expressed in exact form unless the problem specifies a decimal approximation. The integral from 1 to infinity of 1 over x squared, for example, converges to exactly 1, and the candidate is expected to write 1, not 1.000. The integral from 0 to 1 of 1 over the square root of 1 minus x squared equals pi over 2, and the candidate is expected to write pi over 2, not 1.5708. A common error on the FRQ is to round the final answer because the calculator is in normal display mode, and the scoring reader does not award the numerical row to a decimal approximation of an exact value. Candidates should put the calculator in exact mode for the final evaluation, and they should write the answer as a fraction, an integer, or a multiple of pi, depending on the form the antiderivative produced.
Direct comparison and the BC-only comparison test row
The comparison test for improper integrals is a BC-only skill, and it appears on roughly a third of recent FRQs that involve an improper integral. The test has two directions. If a candidate can show that a complicated integrand is bounded above by a simpler integrand whose improper integral converges, then the simpler integral dominates the complicated one and the complicated integral converges. If the candidate can show the opposite, that the complicated integrand is bounded below by a simpler integrand whose improper integral diverges, then the simpler integral forces the complicated one to diverge. Both directions earn the convergence row, but the work required to set up the comparison is the scoring challenge.
For a problem like "does the integral from 1 to infinity of 1 over x squared plus sin squared x converge," the candidate is expected to bound the integrand. The cleanest move is to note that sin squared x is between 0 and 1, so the denominator is between x squared and x squared plus 1, and the integrand is between 0 and 1 over x squared. The integral of 1 over x squared converges, so the original integral converges by direct comparison. The scoring reader is looking for the inequality on the page, the statement that the comparison integral converges, and the conclusion that the original integral converges. A candidate who writes the inequality without the conclusion loses the final row, and a candidate who writes the conclusion without the inequality loses the setup row.
For divergence, the limit comparison test is often cleaner than direct comparison, especially for rational functions. The integral from 2 to infinity of 1 over x minus 1 over x squared behaves like the integral of 1 over x for large x, and the limit comparison ratio of the two integrands approaches 1, a finite nonzero constant. Since the comparison integral diverges, the original integral diverges. The limit comparison test requires the candidate to compute the limit of the ratio, declare it finite and nonzero, and then invoke the convergence of the comparison integral. On a BC exam, the limit comparison test is a three-line argument, and it is one of the more efficient ways to score the convergence row on a problem where a direct antiderivative is unavailable.
When the comparison test is the only path
Some integrands simply do not have closed-form antiderivatives. The integral from 1 to infinity of 1 over 1 plus x to the x power is a famous example, and the only way to score the convergence row is by comparison. For a BC candidate, the recognition that a problem is a comparison problem rather than an antiderivative problem is itself a scoring decision: spending five minutes on a u-substitution that does not work is five minutes the candidate does not have. The first thirty seconds of a comparison problem should be spent inspecting the integrand for a known lower or upper bound, and only if that fails should the candidate attempt a direct antiderivative. This triage decision is part of the BC syllabus, and the FRQ is designed to reward it.
Common pitfalls and how to avoid them on the AP Calculus BC FRQ
The first pitfall is the missing limit wrapper. The candidate computes an antiderivative and evaluates across the infinite or unbounded bound without writing the limit first. The scoring reader cannot award the setup row, and the convergence row may also fail because the candidate has not established that the limit exists. The fix is mechanical: every improper integral, even a trivially convergent one, must be rewritten as a limit before any other work begins. Candidates who have drilled this step on homework problems rarely lose the row on the exam.
The second pitfall is the missing split on an interior discontinuity. The candidate writes a single integral across an interior asymptote and computes the antiderivative without splitting. The antiderivative may exist, but the integral diverges, and the candidate's claim that the integral converges is wrong. The fix is to inspect the interval for any point where the integrand is undefined or unbounded before computing the antiderivative, and to split at every such point. The split is a one-line addition to the setup, and it can be the difference between a four and a six on a six-point FRQ.
The third pitfall is the misuse of L'Hospital's rule. The candidate applies L'Hospital to an expression that is not in 0/0 or infinity/infinity form, or applies it to an antiderivative that has a definite value at the bound. The scoring reader cannot award the L'Hospital row because the prerequisite form is not met. The fix is to check the form of the limit before applying L'Hospital, and to skip the step entirely if the form is not indeterminate. A direct substitution or a comparison argument is almost always available as a backup, and candidates who try L'Hospital as a last resort frequently lose the row.
The fourth pitfall is the rounded final answer. The candidate produces a decimal approximation of an exact value because the calculator is in normal display mode. The numerical row is lost because the answer is not in the expected form. The fix is to set the calculator to exact mode, or to write the answer as a fraction by hand, before moving on to the next problem. The numerical row is often the last row the candidate writes, and it is the easiest row to lose to a calculator setting that can be changed in three seconds.
Worked example: a six-row improper integral FRQ
Consider a representative BC FRQ: evaluate the integral from 0 to infinity of x times e to the negative x dx. The setup row is the limit, written as lim as b approaches infinity of the integral from 0 to b of x e to the negative x dx. The antiderivative row is integration by parts, producing negative x minus 1 times e to the negative x, or any equivalent antiderivative. The evaluation row is the antiderivative at b minus the antiderivative at 0, which simplifies to negative b minus 1 times e to the negative b plus 1. The limit row is the limit as b approaches infinity of that expression, which is an infinity/infinity form that resolves to 0 by L'Hospital. The convergence row is the statement that the limit exists and equals 0. The numerical row is the final answer, which is 1. The total is six rows for six points, and each row corresponds to one line of work on the page.
Now consider a harder example: does the integral from 0 to 1 of ln x dx converge, and if so, to what value? The setup row is lim as a approaches 0 from the right of the integral from a to 1 of ln x dx. The integrand is unbounded near 0, so the one-sided direction must appear on the page. The antiderivative row is integration by parts, producing x ln x minus x. The evaluation row is the antiderivative at 1 minus the antiderivative at a, which simplifies to negative 1 plus a minus a ln a. The limit row is the limit as a approaches 0 from the right of a minus a ln a, which is 0 by a standard result or by L'Hospital. The convergence row states that the limit exists and equals negative 1. The numerical row is the final answer, which is exactly negative 1. This problem is a Type 2 improper integral, and the L'Hospital step is on a ln-type limit rather than a rational-function limit, so the candidate must justify the indeterminate form carefully.
Practice pacing and how to budget the FRQ clock for improper integrals
On the AP Calculus BC exam, the Free Response section gives the candidate 1 hour and 30 minutes for six questions, and improper integrals typically appear on one of the calculator-allowed or one of the calculator-not-allowed problems. The recommended pacing is roughly 15 minutes per FRQ, and an improper integral problem will consume the full 15 minutes if the candidate is not careful. The first two minutes should be spent on identification: which type of improper integral, which bound is infinite or unbounded, where to split if the discontinuity is interior. The next three minutes should be spent on the limit setup and the antiderivative. The next three minutes should be spent on the limit evaluation. The next two minutes should be spent on the convergence statement and the numerical answer. The final five minutes should be spent on checking the work and converting the calculator to exact mode.
For candidates preparing for the exam, the highest-leverage practice is to drill the setup rows. The setup row is the row that distinguishes a 5 from a 6 on most improper integral problems, and it is the row that students most often lose because they skip it in their head and forget to write it on the page. Twenty minutes of practice writing the setup row on a dozen different improper integrals, including Type 1 with infinite bounds, Type 2 with interior discontinuities, and the combined shape, will produce a robust habit that survives the time pressure of the FRQ. The antiderivative row, the evaluation row, and the numerical row are all dependent on the setup row, and they cannot earn full credit if the setup is missing or wrong.
Conclusion and next steps for BC candidates
Improper integrals on the AP Calculus BC exam are graded row by row, and the rows are visible on the page if the candidate knows what to write. The setup row requires a limit, the antiderivative row requires a correct integration technique, the evaluation row requires substitution at a symbolic bound, the limit row requires a finite limit, the convergence row requires the correct vocabulary, and the numerical row requires an exact value. Candidates who drill all six rows on a representative set of problems, including Type 1, Type 2, and combined shapes, can score the full six points with relatively little memorisation. The most common errors are the missing limit, the missing split, the unjustified L'Hospital, and the rounded final answer, and each of these errors can be prevented with a one-second check at the end of the problem. AP Courses' one-to-one AP Calculus BC programme scores each student's improper integral FRQ attempts against the rubric, identifies which of the six rows is consistently lost, and turns the row-by-row scoring into a concrete preparation plan that targets the highest-leverage error first.