Treating an integral of velocity as distance traveled. If velocity changes sign, $\int v(t)\,dt$ gives displacement, and the trips forward and back cancel. For total distance, integrate speed: $\int |v(t)|\,dt$. On a calculator question you can type the absolute value in directly. When you work by hand, find where $v = 0$, split the interval at those times, and add the absolute value of each piece.
Reporting net change when the question asks for an amount. Students compute $\int (\text{in} - \text{out})\,dt$ and then stop, which leaves out the starting value entirely. An amount at time $t$ is the starting amount plus the net change: $A(t) = A(0) + \int_0^t \big(R_{in}(s) - R_{out}(s)\big)\,ds$. Write that whole line out on the page so the reader can see every piece and award each point.
Deciding whether an amount is rising from one rate alone, such as checking only that the inflow rate is positive. The amount's rate of change is inflow minus outflow, so evaluate both rates at the given time, subtract, and state the sign as your reason.
Using equal widths in a Riemann or trapezoidal sum when the table's intervals are uneven, or using the wrong endpoints. A left sum on four intervals uses the four left endpoints, not the right ones and not all five table values. Write each width from the table next to its term, because the widths are rarely equal. For a left sum, pair each width with the value at the start of its interval; for a right sum, with the value at the end.
Mixing up the Fundamental Theorem. Students differentiate the integrand of $\frac{d}{dx}\int_a^x f(t)\,dt$, or forget to subtract the value at the lower limit in $\int_a^b f(x)\,dx = F(b) - F(a)$. With $u$-substitution, they keep the old $x$-limits after switching to $u$. The derivative of $\int_a^x f(t)\,dt$ is $f(x)$. If the upper limit is $g(x)$, the derivative is $f(g(x))\cdot g'(x)$: evaluate the integrand at $g(x)$, then multiply by $g'(x)$. For a definite integral, always compute $F(b) - F(a)$, even when $F(a)$ looks like it should be 0. After a substitution, convert the limits to $u$-values.
Botching a separable differential equation or an Euler step. Students add $+C$ only after solving for $y$, plug the initial condition in at the very end, or reuse the starting slope for every Euler step instead of recomputing it. Separate the variables, integrate both sides, and write $+C$ the moment you integrate. Substitute the initial condition right away to find $C$, and only then solve for $y$. In Euler's method, each step computes a new slope from the current point, $y_{new} = y_{old} + \frac{dy}{dx}\big|_{(x_{old},\,y_{old})}\,\Delta x$, so build a small table with one row per step.
Adding velocity components to get speed, or finding a parametric position by integrating alone. Writing speed as $x'(t) + y'(t)$ is wrong, and so is reporting $\int_0^t y'(s)\,ds$ as the $y$-coordinate. Speed is the length of the velocity vector, $\sqrt{(x'(t))^2 + (y'(t))^2}$, and distance traveled is the integral of that speed. Work out position one component at a time: $x(t) = x(0) + \int_0^t x'(s)\,ds$, and the same for $y$. Left or right depends on the sign of $x'(t)$, not on the slope $dy/dx$.
Dropping the $\frac{1}{2}$ in polar area, or setting up the area between two polar curves as $\frac{1}{2}\int (r_1 - r_2)^2\,d\theta$ with guessed limits. Polar area is $\frac{1}{2}\int_\alpha^\beta r^2\,d\theta$, because each thin slice is a sector, not a rectangle. Between two curves, square each radius first: $\frac{1}{2}\int_\alpha^\beta \big(r_{outer}^2 - r_{inner}^2\big)\,d\theta$. Get the limits by solving $r_1(\theta) = r_2(\theta)$, and check which curve is outside on that interval.
Using the wrong derivative in the Lagrange error bound, or taking the maximum of that derivative over the wrong interval. A common slip is bounding a degree-3 polynomial's error with $f'''$ and $3!$. For a degree-$n$ Taylor polynomial centered at $c$, the error at $x$ is at most $\frac{\max|f^{(n+1)}(z)|}{(n+1)!}|x - c|^{n+1}$, where the max is over $z$ between $c$ and $x$. A degree-3 polynomial therefore uses the fourth derivative and $4!$. Write down the maximum you use and the interval it covers, then compare the bound with the target.
Skipping the justification on series questions: forgetting to test endpoints, misreading the signs at $x = -3$ versus $x = 3$, or stating an alternating series error bound without its conditions. Substitute each endpoint into the original series and simplify the signs before you name a test. For the alternating series error bound, write that the terms alternate, decrease in absolute value and approach 0. Only then compare the first unused term with the target.