Sorry, but there isn't enough information here to know what could be happening.  What
is your command-line?  Can you run with

MKDEBUG=1 mk-query-digest ......

And see if the output just before the error shows you anything useful?

I've found a way to reproduce this, there is no need to respond.  Using 

mk-query-digest  Ver @VERSION@ Distrib @DISTRIB@ Changeset 5021

I can do this:

baron@kanga:~/etc/maatkit/common/t/samples$ for a in slow*; do mk-query-digest
--group-by distill $a >/dev/null; echo ============================$a; done
============================slow001.txt
============================slow002.txt
============================slow003.txt
============================slow004.txt
============================slow005.txt
Use of uninitialized value $num in numeric ge (>=) at /home/baron/bin/mk-query-digest
line 1369.
============================slow006-session-1.txt
Use of uninitialized value $num in numeric ge (>=) at /home/baron/bin/mk-query-digest
line 1369.
============================slow006-session-2.txt
Use of uninitialized value $num in numeric ge (>=) at /home/baron/bin/mk-query-digest
line 1369.
============================slow006-session-3.txt
============================slow006.txt
============================slow007.txt
============================slow008.txt
============================slow009.txt
============================slow010.txt
============================slow011.txt
============================slow012.txt
============================slow013.txt
============================slow014.txt
============================slow015.txt
============================slow016.txt
============================slow017.txt
============================slow018.txt
============================slow019.txt
============================slow020.txt
============================slow021.txt
============================slow022.txt
============================slow023.txt
============================slow024.txt
============================slow025.txt
============================slow026.txt
============================slow027.txt
============================slow028.txt
============================slow029.txt
============================slow030.txt
============================slow031.txt
============================slow032-rewritten.txt
============================slow032.txt
============================slow033.txt
============================slow034.txt
============================slow035.txt
============================slow036.txt
============================slow037.txt
============================slow038.txt
============================slowlogwriter001.txt

Well what do you know: those logs that fail aren't tested in mk-query-digest.t.

Well what do you know: those logs that fail aren't tested in mk-query-digest.t.

Ah that's why: slow006-session-* aren't slow logs, they're files for mk-log-player. 
So we still need a test case for this.

ufaria, can you provide a test case for this or a sample of a log that reproduces the
error?

Summary was "problem on report in mk-query-digest".

ufaria, can you provide a sample of a log that reproduces this error?

I'm going to close this as NonReproducible.  If anyone can provide a way to reproduce
this error, we'll reopen and fix this issue.