The Three Substitution Patterns
Trig substitution only works cleanly for three forms. If your integral doesn't match one of these, stop and look for a different method. The forms are: sqrt(a² - x²), sqrt(x² - a²), and sqrt(x² + a²). Each one maps to a specific substitution based on Pythagorean identities. For sqrt(a² - x²), you substitute x = a sin(). This triggers 1 - sin²() = cos²() and the radical disappears. For sqrt(x² - a²), you use x = a sec(), which uses sec²() - 1 = tan²(). For sqrt(x² + a²), you use x = a tan(), which uses 1 + tan²() = sec²(). These aren't arbitrary. The substitutions exist because the corresponding Pythagorean identities produce perfect squares that cancel the radical entirely.
Trig Substitution Cheat Sheet
Case 1: sqrt(a² - x²) — substitute x = a sin(), dx = a cos() d, the radical becomes a cos(). Reference triangle: angle , opposite side x, hypotenuse a, adjacent side sqrt(a² - x²). Case 2: sqrt(x² - a²) — substitute x = a sec(), dx = a sec() tan() d, the radical becomes a tan(). Reference triangle: angle , adjacent side a, hypotenuse x, opposite side sqrt(x² - a²). Case 3: sqrt(x² + a²) — substitute x = a tan(), dx = a sec²() d, the radical becomes a sec(). Reference triangle: angle , opposite side x, adjacent side a, hypotenuse sqrt(x² + a²).
The cheat sheet you'll find online usually shows these three cases in a table with the substitution, the differential, the simplified radical, and the triangle. That's all you really need to carry with you. Everything else is applying those entries to whatever the integral throws at you.
Get the Full Details
How It Actually Works in Practice
Here's a concrete example. Consider the integral of x² / sqrt(9 - x²) dx from -3 to 3. You identify the form sqrt(a² - x²) with a = 3. Set x = 3 sin(), which means dx = 3 cos() d. The radical simplifies to 3 cos(). Your new integral in is (9 sin²()) / (3 cos()) times 3 cos() d. The cos() terms cancel and you're left with 9 sin²() d. Use the power-reduction identity sin²() = (1 - cos(2))/2, integrate to get (9/2)( - sin(2)/2), then convert back to x using the reference triangle. Since sin() = x/3, you get = arcsin(x/3). The sin(2) term expands to 2 sin() cos() = 2(x/3)(sqrt(9-x²)/3). The full antiderivative is (9/2) arcsin(x/3) - (x/2) sqrt(9-x²) + C. The hard part isn't the substitution itself. It's managing the back-conversion and the algebra that happens after the integral evaluates. Most students mess up at the stage where they have to express sin(2) or cos(2) in terms of x. Drawing the reference triangle once and labeling all three sides prevents errors there. Mark , mark the opposite and adjacent sides based on your substitution, and use Pythagoras for the third side. That triangle becomes your conversion key for every trig function you encounter after integration. When you're evaluating definite integrals, you have a choice: convert the bounds to values before integrating, or keep x bounds and convert the antiderivative back. Converting bounds early is faster but only works cleanly when the bounds correspond to standard angles. For sqrt(9 - x²) with bounds -3 to 3, the bounds become -/2 to /2, which is straightforward. But with non-standard bounds, keeping everything in x until the end is less error-prone.
The Edge Case I Keep Running Into
There's a specific situation where trig substitution looks like it should work but quietly fails unless you handle it correctly. Consider an integral with sqrt(x² + 4x + 13) in the denominator. The quadratic doesn't factor, and it doesn't immediately match any of the three standard forms. You complete the square and get sqrt((x + 2)² + 9). Now it matches the x² + a² pattern with a shift. You set u = x + 2 and u = 3 tan(). The integral becomes a sec³() form, which requires the reduction formula or integration by parts to evaluate. The sec³ integral gives you (1/2)sec()tan() + (1/2)ln|sec() + tan()|. Then you convert back through u to x. The trap here is forgetting to shift the bounds or the variable when converting back. I've seen this cost full credit on exams because the antiderivative in was correct but the final expression in x had the wrong argument inside the logarithm. The workaround is simple but easy to skip: after you finish everything in , write down u = 3 tan() and x = u - 2 as separate notes before you start substituting. It takes thirty seconds and prevents the kind of error where you accidentally write ln|x/3 + sqrt(x²+4x+13)/3| instead of ln|(x+2)/3 + sqrt((x+2)²+9)/3|. The logarithm simplifies to the same value numerically, but the unsimplified form is what the grader is looking for, and missing the shift makes it look wrong on paper.
Common Pitfalls That Waste Time
The most frequent mistake is treating the radical as if it simplifies to the adjacent side without checking which substitution you're using. If you picked x = a sec() for a sqrt(x² - a²) problem, the radical is a tan(), not a sec(). Mixing up which trig function the radical becomes is the fastest way to get a wrong answer that still looks plausible. Another issue is forgetting that the differential dx brings its own trig factors. With x = a tan(), dx = a sec²() d. That sec²() multiplies everything else in the integrand. Students often substitute the radical correctly and then forget the extra sec²() from dx. The integral ends up with one power of secant too few, and the whole evaluation goes off track. Absolutely value signs on secant and tangent also cause problems. When you simplify sqrt(tan²()) to tan(), you're implicitly assuming tan() is non-negative, which requires restricting to the appropriate branch. For x = a sec(), the standard restriction is in [0, /2) union [, 3/2), which means sec() and tan() can have different signs depending on the quadrant. In practice, most textbook problems assume the positive branch, but if your bounds or the problem structure push into a region where tan() is negative, dropping the absolute value changes the sign of your answer.
When Trig Substitution Is the Wrong Tool
Trig substitution breaks down or becomes inefficient in several scenarios. If your integrand contains sqrt(x + 1), no trig substitution will simplify this. The form doesn't match any Pythagorean identity. You'd need elliptic integrals or a completely different approach. If you have a rational function where the denominator factors into distinct linear terms, partial fractions is faster and produces a cleaner answer. Trig substitution can technically handle some rational expressions, but it usually produces logarithmic and inverse trig terms that partial fractions gives directly. Hyperbolic substitution is worth knowing as an alternative for the sqrt(x² + a²) and sqrt(x² - a²) cases. Setting x = a sinh() for sqrt(x² + a²) avoids secants and tangents entirely, and the identity 1 + sinh²() = cosh²() works just as cleanly. Some students find hyperbolic substitution produces simpler intermediate steps because cosh() is always positive, eliminating the absolute value headaches. The tradeoff is that hyperbolic functions are less familiar to most students, and the back-substitution requires inverse hyperbolic notation unless you convert to logarithmic form. There's also a limit to how far trig substitution stretches. It handles integrals where the radical is the dominant difficulty. Once you add additional complex factors like e^x or ln(x) multiplied into the integrand, the method becomes impractical. The substitution simplifies the radical but leaves the other factor untouched, and you're often stuck with an integral that's harder than what you started with.
Building Your Own Reference
The cheat sheet that actually helps during an exam is the one you write yourself. Copy the three cases above into a compact table. Include the substitution, the differential, the simplified radical, and a small triangle diagram for each. Draw the triangles. Writing them out once locks in the side relationships better than memorizing formulas. When you're working a problem under time pressure, you won't have the mental bandwidth to reconstruct which side is opposite and which is adjacent from a verbal description. A quick glance at a drawn triangle solves that in two seconds. I keep a one-page version clipped to my notebook. It's not elegant, but it covers every case I encounter in a standard calculus sequence and takes about twenty seconds to scan when I'm stuck on an integral. The rest is just pattern recognition — seeing sqrt(something² minus x²) and immediately knowing to reach for sine substitution without thinking through the derivation each time.
