As Sergey mentioned using a schema is ideal. If the data comes from a .NET dataset export the dataset with a schema which is the easiest.

If you don't have a schema you can pre-create the cursor to import to and XMLTOCURSOR() can infer the schema based on the cursor's structure. The following works:

TEXT TO lcXmL NOSHOW
<DataBase>
 <ConnectionInfo>
  <DBKey>1</DBKey>
  <String>SQLNCLI10</String> 
  </ConnectionInfo>
</DataBase>
ENDTEXT

CREATE CURSOR TData (DbKey n(5), String c(20))

XMLTOCURSOR(lcXml,"TData",8192)

BROWSE

+++ Rick ---

>Anyone can explain why I am getting a different result using the following expression:
>
>

>cFileName = "test.xml"
>XMLTOCURSOR(cFileName,"alias_name", 512)
>

>
>The content of text.xml looks like following:
>

><DataBase>
> <ConnectionInfo>
>  <DBKey>1</DBKey>
>  <String>SQLNCLI10</String> 
>  </ConnectionInfo>
> <ConnectionInfo>
>  <DBKey>2</DBKey>
>  <String>SQLNCLI10</String> 
>  </ConnectionInfo>
></DataBase>
>

>
>The above XML file produces a cursor with the column DBKey is number, 1 and 2. This is what I expect.
>
>But if I have only one element ConnectionInfo as following:
>

><DataBase>
> <ConnectionInfo>
>  <DBKey>1</DBKey>
>  <String>SQLNCLI10</String> 
>  </ConnectionInfo>
></DataBase>
>

>
>Then the above test.xml produces a cursor where the column DBKey is .T.
>
>Why? Could someone explain what I am missing?