I am just curious why do you need such small reaction volume? There are some major problems of using tiny reaction volumes in PCR:

First, there will always be some level of evaporation in any PCR without using mineral oil overlay, even with heated lid. As the reaction goes down, even tiny amount of evaporation will affect the PCR efficiency significantly.

Second, using small PCR reaction volume will make alloquoting reagents accurately really difficult. Any small variations with your pipetting will also make your PCR efficiency fluctuate.

People in our lab routinely use 10 ul reactions for real-time pcr but there is a reluctance to go lower. I am sure you can, you just may have to accept a higher "noise" level.

We routinely do 10ul rxns for standard and qPCR, no problem at all. We dont go lower because of the error around pipetting smaller volumes, but I guess it probably would be fine.

We are planning to do a three-step reaction starting with RT followed by a pre-amp PCR and then real-time PCR. This is for a highly multiplexed PCR. The preamp without amplification bias is based on a proprietary chemistry @ 10.66 CAD. It's the cost of the pre-amp step at recommended 20 ul that makes me want to try to reduce reaction volumes to a minimum. I'm expecting 10 ul to be fine but was curious if others had hard data on comparison of standard PCR with lower reaction volumes in 0.2 ml reaction tubes.

In our lab, 10uL is fine. We don't go lower than that except for genotyping (6.25uL).

This is totally anecdotal, but our Invitrogen rep said they had validated their qPCR master mixes down to 1 ul. This very well could have been in a 384 well plate though, so take it with a grain of salt.

An invitrogen rep has told me the same thing too. Biorad's new droplet digital PCR does it in nanoliter volumes, but obviously you need their special machine to do so. Our Biorad reps tell me you can 5ul in a 384 well or bigger plate, but that you really need a robot to maintain pipetting accuracy in such a small volume.

As with other people, I have seen 5 ul recommended and routinely do 10 ul PCRs, but we also do 3 ul primer extensions (not as hot as PCR, but can be 12 h at 65C). You do get evaporation, but you do get liquid left!

Also not in 0.2 ml tubes, but rather hard-shell 96-well plates with auto-sealing metal lids.

I have done 10 uL RT-PCR in 200 uL tubes sealed with an adhesive film (BioRad) and saw no significant loss in volume, so 5 uL should work fine, but I think how well your tubes seal is going to be a big factor. Why not make up a 20 uL reaction and aliquot it out in 2.5/5/10 uL volumes and run the reaction to see? If you make a master mix and then use a calibrated P10 or P2 to add the template (at least 1-2 uL) you should have good reproducibility. If you use ART barrier tips with a P10 there is a 2 uL mark on the tip to help in accuracy when adding the template. I use the Eppendorf Reference pipettes for applications like this.

DuPont/Pioneer Hybred routinely uses 3 uL reaction volumes in a 1536-well format. But that is big-business for you. The maize genome is well-wrought out at this point, and they have a perfectly-stable single reference gene to which to normalize everything.

Personally, in agreement with Gary Laco, I think 10 or 15 uL (for 200 uL tubes) is still the way to go with 'academic' samples. I routinely use 20 uL rxn sizes for 'normal' RT-qPCR/qPCR, but use 15 uL sizes for LCM-coupled RT-qPCR.

4.8 uL sample size in 20 uL rxns, and 3.6 uL sample size in the 15 uL rxns.

Maybe too late, but I'd really like to share my experience on this question. Recently, I tried using 5ul and 10ul PCR volume followed by seamless cloning, it failed with the 5ul group, and product from 10ul group worked well. So I would suggest to do PCR no less than 10ul.