What's a subquery?

        7 votes: *****     14,971 views      No comments
by Allen Browne, 23 March 2007    (for ALL VERSIONS of Access)

Microsoft Access Tips for Serious Users

Provided by Allen Browne, March 2007. Updated February 2009.

Subquery basics

Discovering subqueries is one of those "Eureka!" moments. A new landscape opens in front of you, and you can do really useful things such as:

  • Read a value from the previous or next record in a table.

  • Select just the TOP (or most recent) 5 scores per client.

  • Choose the people who have not paid/ordered/enrolled in a period.

  • Express a value as a percentage of the total.

  • Avoid inflated totals where a record is repeated (due to multiple related records.)

  • Filter or calculate values from other tables that are not even in the query.


What is a subquery?

The SELECT query statement

This example shows basic SQL syntax.
It returns 3 fields from 1 table, applies criteria, and sorts the results:

SELECT CompanyID, Company, City
FROM Table1
WHERE (City = "Springfield")
ORDER BY Company;

The clauses must be in the right order. Line endings and brackets are optional.

A subquery is a SELECT query statement inside another query.

As you drag fields and type expressions in query design, Access writes a sentence describing what you asked for. The statement is in SQL (see'quell) - Structured Query Language - the most common relational database language, also used by MySQL, SQL Server, Oracle, DB2, FoxPro, dBase, and others.

If SQL is a foreign language, you can mock up a query like the subquery you need, switch it to SQL View, copy, and paste into SQL View in your main query. There will be some tidying up to do, but that's the simplest way to create a subquery.

Subquery examples

The best way to grasp subqueries is to look at examples of how to use them.

Identifying what is NOT there

A sales rep. wants to hound customers who have not placed any orders in the last 90 days:

SELECT Customers.ID, Customers.Company
FROM Customers
   (SELECT Orders.OrderID                         
   FROM Orders                                    
   WHERE Orders.CustomerID = Customers.CustomerID 
     AND Orders.OrderDate > Date() - 90)          ;

The main query selects two fields (ID and Company) from the Customers table. It is limited by the WHERE clause, which contains the subquery.

The subquery (everything inside the brackets) selects Order ID from the Orders table, limited by two criteria: it has to be the same customer as the one being considered in the main query, and the Order Date has to be in the last 90 days.

When the main query runs, Access examines each record in the Customers table. To decide whether to include the customer, it runs the subquery. The subquery finds any orders for that customer in the period. If it finds any, the customer is excluded by the NOT EXISTS.

Points to note:

  • The subquery goes in brackets, without a semicolon of its own.

  • The Orders table is not even in the main query. Subqueries are ideal for querying about data in other tables.

  • The subquery does not have the Customers table in its FROM clause, yet it can refer to values in the main query.

  • Subqueries are useful for answering questions about what data exists or does not exist in a related table.


Get the value in another record

Periodically, they read the meter at your house, and send a bill for the number of units used since the previous reading. The previous reading is a different record in the same table. How can they query that?

A subquery can read another record in the same table, like this:

SELECT MeterReading.ID,
   (SELECT TOP 1 Dupe.MeterValue                 
   FROM MeterReading AS Dupe                     
   WHERE Dupe.AddressID = MeterReading.AddressID 
     AND Dupe.ReadDate < MeterReading.ReadDate   
   ORDER BY Dupe.ReadDate DESC, Dupe.ID)          AS PriorValue
FROM MeterReading;

The main query here contains 4 fields: the primary key, the reading date, the meter value at that date, and a fourth field that is the value returned from the subquery.

The subquery returns just one meter reading (TOP 1.) The WHERE clause limits it to the same address, and a previous date. The ORDER BY clause sorts by descending date, so the most recent record will be the first one.

Points to note:

  • Since there are two copies of the same table, you must alias one of them. The example uses Dupe for the duplicate table, but any name will do.

  • If the main query displays the result, the subquery must return a single value only. You get this error if it returns multiple values:
        At most one record can be returned by this subquery.

  • Even though we asked for TOP 1, Access will return multiple records if there is a tie, e.g. if there were two meter readings on the same date. Include the primary key in the ORDER BY clause to ensure it can decide which one to return if there are equal values.

  • The main query will be read-only (not editable.) That is always the case when the subquery shows a value in the main query (i.e. when the subquery is in the SELECT clause of the main query.)

TOP n records per group

You want the three most recent orders for each client. Use a subquery to select the 3 top orders per client, and use it to limit which orders are selected in the main query:

SELECT Orders.CustomerID, Orders.OrderDate, Orders.OrderID
FROM Orders
WHERE Orders.OrderID IN
   (SELECT TOP 3 OrderID                            
   FROM Orders AS Dupe                              
   WHERE Dupe.CustomerID = Orders.CustomerID        
   ORDER BY Dupe.OrderDate DESC, Dupe.OrderID DESC) 
ORDER BY Orders.CustomerID, Orders.OrderDate, Orders.OrderID;

Points to note:

  • Since we have two copies of the same table, we need the alias.

  • Like EXISTS in the first example above, there is no problem with the subquery returning multiple records. The main query does not have to show any value from the subquery.

  • Adding the primary key field to the ORDER BY clause differentiates between tied values.

Year to date

A Totals query easily gives you a total for the current month, but to get a year-to-date total or a total from the same month last year means another calculation from the same table but for a different period. A subquery is ideal for this purpose.

SELECT Year([Orders].[OrderDate]) AS TheYear,
Month([Orders].[OrderDate]) AS TheMonth,
Sum([Order Details].[Quantity]*[Order Details].[UnitPrice]) AS MonthAmount,
   (SELECT Sum(OD.Quantity * OD.UnitPrice) AS YTD                                
   FROM Orders AS A INNER JOIN [Order Details] AS OD ON A.OrderID = OD.OrderID   
   WHERE A.OrderDate >= DateSerial(Year([Orders].[OrderDate]),1,1)               
     AND A.OrderDate < DateSerial(Year([Orders].[OrderDate]),                    
       Month([Orders].[OrderDate]) + 1, 1))                                       AS YTDAmount
FROM Orders INNER JOIN [Order Details] ON Orders.OrderID = [Order Details].OrderID
GROUP BY Year([Orders].[OrderDate]), Month([Orders].[OrderDate]);

Points to note:

  • The subquery uses the same tables, so aliases them as A (for Orders) and OD (for Order Details.)

  • The date criteria are designed so you can easily modify them for financial years rather than calendar years.

  • Even with several thousand records in Order Details, the query runs instantaneously.

Delete unmatched records

The Unmatched Query Wizard (first dialog when you create a new query) can help you identify records in one table that have no records in another. But if you try to delete the unmatched records, Access may respond with, Could not delete from specified tables.

An alternative approach is to use a subquery to identify the records in the related table that have no match in the main table. This example deletes any records in tblInvoice that have no matching record in the tblInvoiceDetail table:

DELETE FROM tblInvoice
   (SELECT InvoiceID                                        
   FROM tblInvoiceDetail                                    
   WHERE tblInvoiceDetail.InvoiceID = tblInvoice.InvoiceID);

Delete duplicate records

This example uses a subquery to de-duplicate a table. "Duplicate" is defined as records that have the same values in Surname and FirstName. We keep the one that has the lowest primary key value (field ID.)

WHERE ID <>  (SELECT Min(ID) AS MinOfID FROM Table1 AS Dupe      
   WHERE (Dupe.Surname = Table1.Surname)                         
   AND (Dupe.FirstName = Table1.FirstName));                     

Nulls don't match each other, so if you want to treat pairs of Nulls as duplicates, use this approach:

WHERE ID <>  (SELECT Min(ID) AS MinOfID FROM Table1 AS Dupe      
   WHERE ((Dupe.Surname = Table1.Surname)                        
     OR (Dupe.Surname Is Null AND Table1.Surname Is Null))       
   AND ((Dupe.FirstName = Table1.FirstName)                      
     OR (Dupe.FirstName Is Null AND Table1.FirstName Is Null))); 

Aggregation: Counts and totals

Instead of creating a query into another query, you can summarize data with a subquery.

This example works with Northwind, to show how many distinct clients bought each product:

SELECT Products.ProductID, Products.ProductName, Count(Q.CustomerID) AS HowManyCustomers
 FROM Orders INNER JOIN [Order Details] ON Orders.OrderID = [Order Details].OrderID) AS Q
INNER JOIN Products ON Q.ProductID = Products.ProductID
GROUP BY Products.ProductID, Products.ProductName;

Points to note:

  • The subquery is in the FROM clause, where it easily replaces another saved query.
  • The subquery in the FROM clause can return multiple fields.
  • The entire subquery is aliased (as Q in this example), so the main query can refer to (and aggregate) its fields.
  • Requires Access 2000 or later.

Filters and searches

Since subqueries can look up tables that are not in the main query, they are very useful for filtering forms and reports.

A Filter or WhereCondition is just a WHERE clause. A WHERE clause can contain a subquery. So, you can use a subquery to filter a form or report. You now have a way to filter a form or report on fields in tables that are not even in the RecordSource!

In our first example above, the main query used only the Customers table, and the subquery filtered it to those who had no orders in the last 90 days. You could filter the Customers form in exactly the same way:

'Create a subquery as a filter string.
strWhere = "NOT EXISTS (SELECT Orders.OrderID FROM Orders " & _
    "WHERE (Orders.CustomerID = Customers.CustomerID) AND (Orders.OrderDate > Date() - 90))"
'Apply the string as the filter of the form that has only the Customers table.
Forms!Customers.Filter = strWhere
Forms!Cusomters.FilterOn = True
'Or, use the string to filter a report that has only the Customers table.
DoCmd.OpenReport "Customers", acViewPreview, , strWhere

This technique opens the door for writing incredibly powerful searches. Add subqueries to the basic techniques explained in the Search form article, and you can offer a search where the user can select criteria based on any related table in the whole database.

The screenshot below is to whet your appetite for how you can use subqueries. The form is unbound, with each tab collecting criteria that will be applied against related tables. The final RESULTS tab offers to launch several reports which don't even have those tables. It does this by dynamically generating a huge WhereCondition string that consists of several subqueries. The reports are filtered by the subqueries in the string.

Subquery search screenshot

Other examples

These articles also illustrate the use of subqueries:


Now you know how powerful subqueries are, you will also want to know their limitations. Surviving subqueries gives insight to help you trouble-shoot them.

Home Index of tips Top

Rate this article:  Your rating: PoorYour rating: Not so goodYour rating: AverageYour rating: GoodYour rating: Excellent

This is a cached tutorial, reproduced with permission.

Have your say - comment on this article.

What did you think of 'What's a subquery?'?

No comments yet.

Why not be the first to comment on this article?!

Have your say...

E-mail (e-mail address will be kept private)

Comments require approval before being displayed on this page (allow 24 hours).