not reliably -- you can get a good "guess" by seeing the time at the top of the tracefile and subtracting. for example:

alter session set sql_trace=true;
select * from dual d1;
exec dbms_lock.sleep(5)
select * from dual d2;


now the trace file will have:


*** 2005-05-09 10:14:40.662
*** SESSION ID:(11.4450) 2005-05-09 10:14:40.661
APPNAME mod='SQL*Plus' mh=3669949024 act='' ah=4029777240
=====================
PARSING IN CURSOR #3 len=32 dep=0 uid=245 oct=42 lid=245 tim=1089500078771773 hv=3943786303 ad='576def28'
alter session set sql_trace=true
END OF STMT
=====================
PARSING IN CURSOR #3 len=21 dep=0 uid=245 oct=3 lid=245 tim=1089500078773194 hv=430859874 ad='576de3bc'
select * from dual d1
END OF STMT
=====================
PARSING IN CURSOR #3 len=32 dep=0 uid=245 oct=47 lid=245 tim=1089500078883265 hv=1700220278 ad='576dd38c'
BEGIN dbms_lock.sleep(5); END;
END OF STMT
PARSE #3:c=10000,e=101446,p=7,cr=63,cu=0,mis=1,r=0,dep=0,og=4,tim=1089500078883256
EXEC #3:c=0,e=4993747,p=0,cr=0,cu=0,mis=0,r=1,dep=0,og=4,tim=1089500083877160
=====================
PARSING IN CURSOR #3 len=21 dep=0 uid=245 oct=3 lid=245 tim=1089500083880596 hv=379045529 ad='576ec038'
select * from dual d2
END OF STMT


taking the last tim= and the first tim=, we get:

ops$tkyte@ORA9IR2> select (1089500083880596-1089500078771773)/1000000 from dual;

(1089500083880596-1089500078771773)/1000000
-------------------------------------------
5.108823


(would be /1000 in 8i) about 5.1 seconds -- add that to:
*** SESSION ID:(11.4450) 2005-05-09 10:14:40.661

and you get 10:14:45 approximately.


as long as the tim= records are 1/1,000,000 of a second of course. could be platform specific. I was using linux