The basic idea is to set up a parameter as if it were going to be a single value parameter with the topmost row selected by default.  I will walk through the steps of this scenario.

I chose the AdventureWorks database to do this sample and focused on June 2004 Sales to get an easy dataset to work with.

1. Create the basic dataset:

2. Set up the layout for the report:

3. Preview the report with no parameters:

4. Create a new dataset for the territories list:

5. Set up the parameter, TerritoryID, like a normal single-value parameter.  Be sure to set the default value to make it default to the topmost row in the list.  In this case, the topmost is “Australia”, TerritoryID of 9.

6a. Apply the filter to the table.  Notice that I am not editing the dataset, but simply adding a filter to the table properties.  If you are usually the type to edit the dataset (like myself), just hang with me for a second.  Right-click the edge of the table and then choose Properties:

6b.  Modify the table properties to add the filter.  Since you can’t read it all, it says

=Fields!TerritoryID.Value = Parameters!TerritoryID.Value

7. Check the finished result.  We now have a report with a single parameter.  The topmost selection, Australia, is automatically selected and our filter is applied by default.

8. So far, so good–there is nothing extraordinary about what we’ve done so far.  Now, we simple make the parameter multi-value by changing the report parameters and checking the Multi-value box:

9. Don’t rush off just yet, or you’ll see this error message when you choose Preview:

10.  You need to make one small change to the filter.  Open the table properties again, but this time change the operator from equals (=) to IN.  Since the Parameters!TerritoryID.Value now contains a comma separated list, we need to search IN instead of comparing.

11. Now, preview your report.  The left image shows the initial preview, the right image shows the drop-down list blown out, you can see that Select All is selected by default.

One caveat is that performance is not going to be top notch because Reporting Services is filtering on the entire dataset every time.  It would (most likely, depending on your implementation) be faster to filter in the dataset.  However, this approach often requires a lot of development work on the T-SQL side to get it working.  This approach is easy and quick enough for many applications.  Enjoy.

Update: I found a good example of the hard way to do multi-value parameters (by using SQL).  If you scroll down to the second post, there is source code there for converting a delimited string into a table variable.  I have re-formatted the charlist_to_table source code to make it more legible.

I wanted users to use the standard date-picker, but there is no way to format the date inside the query.  My work-around is a classic Reporting Services pattern – create an internal parameter that does the formatting for you.  The format string for VBScript is “yyyy-MM-ddTHH:mm:ss”.

Below is a screenshot of the Report Parameters dialog. I used the following code to just grab today’s date (no time).

     =CDate(CDate(Now()).ToString("MM/dd/yyyy"))

For the second parameter, I create an internal parameter that is dependent on the first.  This will prevent it from showing on the report and it will not be filled in until the start date is chosen.  Default value:

     =CDate(Parameters!StartDate.Value).ToString("yyyy-MM-ddTHH:mm:ss")

I created a test report to show how this works.  Here is the report definition:

Below is the report preview:  You can now use the parameter in your CAML query.

One other note on CAML date fields.  The time is ignored by default.  If you want to include the time, you must specify another attribute.  Example:

<Geq>

<FieldRef Name="LastModified" />

<Value Type="DateTime" IncludeTimeValue="TRUE">@LastModified!</Value>

</Geq>

I found a helpful post on CAML & DateTime that might be useful for you too.