SQL Subquery Practice — 57 Challenges (42 Free)

A subquery as a value or a list — scalar and IN: 36 challenges (25 free)

WHERE salary > (SELECT AVG(salary) FROM employees) — the inner query runs once and returns one value; WHERE customer_id IN (SELECT …) returns a list to match against. The Easy on-ramps are here: A Subquery That Returns One Value, then the same shape on a second schema in Movies Rated Above the Average; then A Subquery That Returns a Set (IN), and Every Film by a Director Who Once Hit 8.5, where the list comes out of the table you are already selecting from. The Medium ones move the scalar into the SELECT list, into HAVING, inside COALESCE; the Hard ones nest it inside a CTE or divide by it for a share. Mind NOT IN against a nullable column — the NOT IN trap.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard.

Correlated subqueries and EXISTS: 11 challenges (9 free)

The inner query references the outer row — WHERE e1.salary > (SELECT AVG(e2.salary) FROM employees e2 WHERE e2.department = e1.department) — so it runs once per row. EXISTS is the correlated form for existence checks, and NOT EXISTS is the anti-join that survives NULLs where NOT IN does not. Start on the two Easy ones: Better Than Its Own Genre, where a single m2.genre = m.genre turns one average into one average per row, and Who Manages Nobody, where the NOT IN version of the same question really does return zero rows. Then Departments With High Earners and EXISTS vs IN; Correlated Subquery: Employees Above Department Median is the Hard one that names the pattern. The anti-join explained has the NOT EXISTS walkthrough.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard.

Derived tables — a subquery in FROM: 18 challenges (13 free)

Aggregate first, then query the result — FROM (SELECT department, AVG(salary) AS avg_salary FROM employees GROUP BY department) d. Start with Filter on an Average You Just Computed, the Easy one — a country average computed in FROM and then filtered with a plain WHERE — then Subquery in FROM (Derived Table) and Above-Average Departments. The Hard ones wrap a window function you cannot filter in the same SELECT — ROW_NUMBER, LAG, a running SUM — and Running Total Revenue and Year-over-Year Growth are the 2 free Hard previews. Three sector ones wrap a UNION ALL. A derived table is the case where a CTE reads better; the CTE tutorial shows the rewrite.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard — previews before Pro.

The subquery is the inside-out spelling; the same problems written top-down with WITH are on the CTE page, and several Hard ones here open with WITH and appear on both. When the choice matters — and when it does not — is on CTE vs subquery vs temp table.