SQL Cheat Sheet: A Complete Reference for SQL Queries, Filtering, Joins, and Data Manipulation

Most database work runs on a small core of commands. Learn roughly forty of them properly, and you can answer almost any question a table will give up. This SQL cheat sheet covers that core, with the traps that catch people out marked as you go. 

Every query follows one order. You SELECT the columns, FROM a table, JOIN anything else you need, then WHERE to filter rows. After that comes GROUP BY to bucket them, HAVING to filter those buckets, then ORDER BY and LIMIT. Learn that sequence and the rest is detail. 

The SQL cheat sheet at a glance

Family	Key words	What it does
Reading data	SELECT, FROM, DISTINCT, AS	Pulls and names columns
Filtering	WHERE, AND, OR, NOT, IN, BETWEEN, LIKE	Narrows the rows
Combining	INNER, LEFT, RIGHT, FULL, CROSS JOIN, UNION	Brings tables together
Summarising	COUNT, SUM, AVG, MIN, MAX, GROUP BY, HAVING	Turns rows into totals
Ordering	ORDER BY, LIMIT, OFFSET	Sorts and trims output
Changing data	INSERT, UPDATE, DELETE, MERGE	Writes to tables
Structure	CREATE, ALTER, DROP, TRUNCATE	Builds and removes tables
Advanced	WITH, ROW_NUMBER, RANK, OVER	Named subqueries and rankings

Key Takeaways: Where filters rows before grouping, and HAVING filters after. Null is not a value and never equals anything, including itself. And LIMIT is the single most portable-looking keyword that is not actually portable. 

Reading and filtering

• SELECT name, price FROM products; pulls two columns 
• SELECT DISTINCT country FROM customers; removes duplicates 
• SELECT price * 0.8 AS sale_price FROM products; calculates and renames 
• WHERE price > 100 AND stock > 0 combines conditions 
• WHERE country IN beats a chain of ORs 
• WHERE created BETWEEN ‘2026-01-01’ AND ‘2026-06-30’ covers a range 
• WHERE name LIKE ‘Sm%’ matches a prefix, and ‘%th%’ matches anywhere 

Two habits save hours. Write conditions against indexed columns where you can, and avoid wrapping a column in a function inside WHERE, since that usually stops the index being used. 

Joins, in one table

Join	Returns	Use it for
INNER JOIN	Only rows matching in both tables	Orders that definitely have a customer
LEFT JOIN	All left rows, matches where they exist	Customers including those with no orders
RIGHT JOIN	All right rows, matches where they exist	Rarely used; rewrite as a LEFT JOIN
FULL OUTER JOIN	Everything from both sides	Reconciling two partial lists
CROSS JOIN	Every combination of both	Building date or size grids
SELF JOIN	A table joined to itself	Staff and their managers

The classic mistake is putting a condition on the right-hand table in WHERE after a LEFT JOIN. That quietly turns it into an inner join and drops the unmatched rows you were trying to keep. Put the condition in the ON clause instead. 

Grouping and totals SQL cheat sheet

• COUNT(*) counts rows, COUNT(column) skips NULLs 
• SUM(total), AVG(total), MIN and MAX do the obvious 
• GROUP BY country buckets rows before the maths 
• HAVING COUNT(*) > 5 filters the buckets afterwards 
• Every non-aggregated column in SELECT must appear in GROUP BY 

Remember the order of operations. WHERE runs first and removes rows, then grouping happens, then HAVING trims the groups. Putting an aggregate in WHERE returns an error in every major database. 

NULL handling, where most bugs live

SQL cheat sheet: NULL means unknown, not zero and not an empty string. Comparing anything to it with an equals sign returns unknown rather than true, which is why WHERE column = NULL silently matches nothing. 

• Use IS NULL and IS NOT NULL, never = NULL 
• COALESCE(nickname, first_name, ‘friend’) returns the first non-null value 
• NULLIF(a, b) returns NULL when the two match, handy for avoiding divide-by-zero 
• Aggregates skip NULLs, so an average of five rows with two NULLs divides by three. 

Writing and changing data

• INSERT INTO products (name, price) VALUES ; 
• Update products Set price = 12.99 WHERE id = 4; 
• DELETE FROM products WHERE stock = 0; 
• TRUNCATE TABLE logs; empties a table far faster than DELETE 

Run every UPDATE and DELETE as a SELECT first with the same WHERE clause. If the row count looks wrong, you have just saved yourself a restore from backup. Wrap anything irreversible in a transaction where your database supports one. Explore our SaaS category for more software reviews, comparisons, and expert guides. 

Window functions worth knowing

These calculate across a set of rows without collapsing them into one, which is what separates a report from a summary. 

• ROW_NUMBER() OVER (ORDER BY total DESC) numbers rows 
• RANK() ties share a position and skip the next; DENSE_RANK() does not skip 
• SUM(total) OVER (PARTITION BY customer_id) gives a per-customer total on every row 
• LAG(total) OVER (ORDER BY date) reads the previous row, useful for growth calculations 

Where the databases disagree

Task	PostgreSQL and MySQL	SQL Server	Oracle
Limit rows	LIMIT 10	SELECT TOP 10	FETCH FIRST 10 ROWS ONLY
Substring	SUBSTRING()	SUBSTRING()	SUBSTR()
Join text	CONCAT() or ||	CONCAT() or +	CONCAT() or ||
Current time	NOW()	GETDATE()	SYSDATE
String quotes	Single only	Single, double for names	Single only

Sorting of NULLs and case sensitivity both depend on collation settings rather than the standard, so test rather than assume when you move a query between systems.

 Conclduion

Pick one table you already know and write five queries against it without looking anything up: a filter, a join, a group, a ranking and an update you never run. Recall beats reading, and the commands stick after roughly three sessions. SQL cheat sheet: If you spend long days at a keyboard, our guides to the right mouse for desk work and a reliable home router are worth a look. For wider reading, we also compare the tech review sites you should follow. 

Apart from that, if you want to read more articles like this, XLOOKUP vs VLOOKUP, then visit our SaaS category

Frequently Asked Questions