>1.) The cause of the index becoming corrupted or out of sync with table data remains unknown.
>
>2.) The table is created just a few lines earlier, and the index is a CDX index tag. As far as I know, CDXs are typically maintained automatically.
>
>3.) Despite the Error 20 being thrown, the output CSV was generated, and several thousand rows were successfully written to the file. Presumably the issue arose when FoxPro encountered the record with the corrupted index entry.
>
>4.) Upon inspecting the rows in the output CSV (both the shortened version before implementing REINDEX and the full one created after adding REINDEX), I haven't identified any anomalous data that would explain the occurrence of the problem.
>
>5.) I still need to examine the source table's data, but I'm currently running a test, so access to the data is temporarily unavailable.
>
>I hope this provides a clearer picture of the situation. If you have any insights or suggestions, please feel free to share.

Did you make sure that the collation on the tag is machine? I've seen weird things happen when it's anything else, and the index expression becomes a value conformant to current codepage's rules, not necessarily the same value that was calculated. Should not happen on character fields, but for any other kind of field, there's a possibility.