Smoke :  smoke testing is a non ecxhaustive software 
testing he we will see the most crucial function of a 
program but need not bother about finer details 

SANITY:
iT IA A NARROW REGRESSION TEST , HERE WE WILL FOCUS ON ONE 
ARE MORE FUNCTIONAL AREA OF APPLICATION 

sANITY TESTING IS NARROW AND DEEP

Smoke Testing: When a build is released to testing 
department, the test engineer would first check for 
availability of all major functionalities in the build.  If 
there is any major functionality missing in the build, the 
tester would reject the build and would ask for a patch. It 
is also called build verification testing or tester 
acceptance testing.

Sanity Testing: In case of sanity testing, a test engineer 
would check for any extra functionalities in the build with 
respect to requirements document. It is also called garbage 
testing.

Smoke testing:
Smoke testing is usually did by the developers,which helps 
them to decide whether the application is ready for 
testing...Whether will give the build for testing like 
that...
Sanity testing:
Sanity testing is done by the testers which helps then to 
decide whether to accept the build or not..Major 
functionalities have been tested here...

Please correct me if it is wrong....

hello anitha ur answer is correct,i learned like this.

Bit confused with above answers.What i have been hearing is

Smoke
1. More of build acceptance testing. 
2. Covers a wider area of functionality at high level.
3. Not in depth testing.
4. Test Results should be published on within hours.

Sanity 
1. Covers a narrow functionality.
2. More in Depth Testing
3. Test Results should be published on within hours.

When will sanity testing been done.
One thought is when some defect get fixed we might require
to perform a more in depth testing of corrected functionality.

Then will regression go.
My understanding is regression can cover more functionalit.
Bcoz regression would be prepared for whole application and
we test engineers can choose which needs be regression tested. 

PLease reply with correct answers. Looks like too many
technical jargons for same testing.

Srijith's Ans is right.
Again with Smoke testing, When a build is received and say
there are 10 tester's who were suppose to test the build,
then a group of 2-3 testers will make a quick round of
testing where the will find commen bugs (say installation
error). Otherwise if 10 testers starts testing and all gets
same error then it would be wastage of time of all 10 testers.

Where Sanity testing will be performed to check major
functionality is working or not for a perticular build.

--
Mayur K