Based on your statement I suppose that your template contains DNA and you should eliminate it. What method do you use to to isolate RNA? I have good experience with kits based on silica columns coupled with on column DNase digestion. What kind of material do you use (cells from culture, serum etc)?

My go-to method for eliminating non-specific amplification is to re-design my primers, but it sounds like that may not be viable for your work.

It's interesting that your positive control is fine, but the other samples are not. If your positive control is a plasmid, phage, or otherwise a piece of non-genomic DNA you may not be seeing the non-specific amplification products simply because there isn't any non-specific DNA to amplify.

If you cannot re-design your primers you could test the usual steps to reduce non-specific amplification in PCR. Some combination of the below:

Increase the annealing temperature

Decrease the annealing time

Decrease the elongation time (match it to your desired product, not longer).

Decrease the number of amplification cycles (will reduce the maximum CT you can detect, though)

ADDED:

Decrease primer concentration

Decrease template concentration

Arkadiusz, I am working with both cell lines and primary cultures. I am using the TRIzol-chloroform method to extract my RNA. I haven't been doing a DNAse digestion step, but don't see a gDNA smear when I run my RNA on a gel and all my primers are junction-spanning. The non-specific peaks I am seeing are before the specific target which means the corresponding amplicons should be smaller in size, which had me thinking it was unlikely to be amplification of genomic DNA. Though I guess you mean that my primers are adhering to and amplifying random short genomic fragments rather than the target gene with intron(s)? I will look into DNAse digestion options that will be compatible with the TRIzol method.

Brian, I am going to try redesigning, but my genes are a multigene family coding for short peptides, meaning there are limited regions to which I can target a primer to amplify one family member versus the others. My positive control is generally cDNA from a previous cell stimulation experiment, or from a tissue which constitutively expresses the target gene. So the RNA extraction and cDNA synthesis has been carried out the same as my experimental samples. I will definitely try adjusting the thermocycling conditions.

Thanks for the help guys!

In my opinion Trizol is not the best way to extract RNA because it is difficult to obtain RNA without any contaminant, particularly traces of DNA. It can be nonvisible on the gel but detectable with PCR. In the first step I suggest you to obtain RNA of better quality (isolated using silica column and on-column DNase digestion). Simultaneously we can consider another possible reasons: what about annealing temperature? Too low promotes non-specific amplification but my experience indicates that too high may also cause the amplification of strange products.

Maybe write also something about RT conditions - an enzyme and temperature.

We combine trizol methode with silica colums. Trizol, gives a higher yield, but the silica colums makes this higher yield more pure to use. I get RNA this way with OD 260/280 of around 2.00. Make sure you eleminate the ethanol as much as possible from the last step.

For me I have good results if I keep my RNA input in the cDNA conversion reaction at 1µg, or at least higher than 0.5µg, but that is stretching for me.

A melt peak before your peak of interest doesn't actually mean this is a shorter fragment. The melting temperature is determined, not only by length, but also the sequence. You can have 2 fragements of 200bp, the one has a Tm of 85°C the other one 79°C. This is dependent on the GC % of your amplicon.

Preforming melt curve analysis is usually sufficient to  to confirm the melt curve analysis by agarose gel electrophoresis. If the NTC contains specific amplicon, it is better to discard all primers, master mix, and water, decontaminate pipettes and work area and start again with fresh stocks of all reagents and primers. If the product is either a non-specific amplicon or primer dimer it may be best to redesign the primers.

Trizol is the best method for RNA extraction except when extracting very high numbers of samples. Of course washing step (2x, 75% EtOH, removing drops) is very important. Always better to do a DNase (liquid better, 30min, +phenolChlo, +Chlo) since DNA conta not necessary visible on a gel.

SYBRgreen is OK. For your QPCR trouble, take a non precious positive sample, dilute in a non precious negative sample (meaning RNA from cells without your target) to obtain a non precious trouble maker sample. Then optimize the QPCR conditions: if you are in a 2 steps, then go to 3 steps and try from 58 to 65C annealing T and see if you find a better condition. You can also use different QPCR mix from other companies that can give very different results especially with trouble maker primer pairs. If nothing is working, then try to redesign at least one of the primers...