A couple more thoughts.

I noticed that you have

HideReferrer    Direct Request

in your configuration. You need to use a dash for this to work:

HideReferrer    -

You can try XML reports - they have pop-up help when you hover your mouse over some columns, like Hits. There are no any translations yet, though, just in English.

Where do yo see that in my confs ? I don't find.
I never want to hide Direct Request...

In the sample configuration you sent me. It's commented out, but I wanted to mention it anyway, in case you end up using it.

ok, thanks.

About the 2 urls i have send to you by mail, i have some difficulty to understand the differences.
The only thing change in the 2 conf is the IgnoreReferrer xxx.com activated in one, not in the other.

The difference in visits in March :
8420 visits (with : IgnoreReferrer xxx.com activated)
47032 visits (with : #IgnoreReferrer xxx.com)

The difference in Direct Requet in the two case :
2,44%
77,19%

How Internal referrers (activation or not) can do that ?
Internal referrer can't do visitors.
Only direct request and other referrers can do visitors...

There is someting i couldn't explain...

(i understand that there is a bug in Direct Request strange increase)

I think we have established the difference in referrer percentages. That is, you reduced the total number of hits by removing the most frequent item, the internal referrer, the remaining counts will yield higher percentages.

Regarding visits. A visit is a sequence of requests not too far apart from each other (<30 min by default). If you filter out some requests, you will change visits. For example, some of the robots may crawl the site in short visits and report a referrer from the end of the previous visit, which would start a new visit with an internal referrer. Such visits will disappear if you filter out internal referrers.

When it comes to visits, it's better to separate robots and spammers from humans. Try adding configuration to identify robots/spammers and you will see more accurate visit stats.

> (i understand that there is a bug in Direct Request strange increase)

I didn't say there's a bug :) just that I don't mind looking into this if you can reproduce it with a fixed set of logs. That is, you will see these counts different if you analyze live logs.