>Is it true ? Is there particular situations where there are more chances to get corrupted indexes ? Is there any ways to avoid corrupted indexes ?

Others have answered your question well - under a solid HW environment, and using good vfp data retrieval practices such as views/buffering, corruption is very infrequent. Here's one addition I swear by, that has reduced my index corruption and other related vfp data problems even further.

When planning your data scheme, create two major server data directories as your basic structure (this is not vital from a Permissions standpoint, but is much easier to maintain) - one directory for user Editable files, and one for ReadOnly files. Throw every possible file into the ReadOnly server directory. This includes most DBCs (and DCT/DCXs), and especially large informational tables with many tags that can be updated by admins "offline". In fact, even plan your tables this way whenever possible: when updates don't need to be instantaneous or admins/update programs can do them. I had a major dropoff off assorted problems since I implemented this technique about 2 years ago (during vfp5 days), and it also beefs up security simultaneously.