>>>>>>>I tried when I used VFP9 but still different number of records. Actually the program does not delete records in this table.
>>>>>>
>>>>>>Have you verified that it's indeed the same folder and same table? What if the users are mapped differently? Try adding something to that folder and then verify if both users will see that same file.
>>>>>
>>>>>I did verify that. When I was connected to one desktop I created a test text file in the folder. Then from the other desktop I could see this file. And there is only one table of this name.
>>>>
>>>>The only other thing I can think of is to get rid of all indexes by using DELETE TAGS ALL (for now) and verify after that.
>>>
>>>I will try this when I connect next time (I don't have a remote connection, I do it via GoToMeeting). But somehow I think what Tamar said could be the cause. Because I reindexed the tables and still had the problem.
>>
>>What is the device that's storing the data files - server computer (Windows or Linux?), workstation, NAS, ?
>>
>>Can you test with real-time antivirus scanning disabled on all machines that are involved?
>
>The data is on the Windows 2008 server (I think, although not 100% sure), virtual server.
>
>I will ask the IT (if they ever call) to disable the AV. But would AV block user from seeing some records?

From the POV of workstations on the network, there is no difference between a virtualized server and a "real" server running on bare metal.

Another type of "virtualization" people talk about is Windows' remapping of certain folders and registry entries in a user's profile. However, files stored on a network share are not subject to this.

Before going any further I would 1,000% confirm you're looking at the same table from both computers. When you've manually opened it in the VFP9 Command window, do DISPLAY STATUS and note the exact path/file name that's open. If that refers to a mapped drive, fire up a CMD window and run NET USE to show the current drive mappings.

As for AV - if you think you've got funky network stuff going down, it's always a good idea to remove realtime scanning from the list of variables.