To add to the above:

part of the problem might be a mismatch between your primer Tms and your PCR annealing temp; that is your annealing temp is < Tm-2C

• PCR success is a function of efficiency and specificity
• Efficiency culminating in abundant copies of your PCR product is related to then proportion of targets your primers bind to; especially during the first 5 cycles
• Specificity is ensuring that your primers bind to just one target site
• Specificity can be verified by subjecting your primers to BLAST analysis and ensuring that you only get 100% match for the full length of each primer against the required target
• https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE=BlastSearch
• In addition, to ensure your primers ONLY bind to this target site you need to optimise your annealing temp against your primer melting temp
• Thus, calculate your primer melting temp using freeware such as IDT primer design tool, e.g.
• https://www.idtdna.com/calc/analyzer
• Alternatively, enter your target sequence and allow software to design primers for you, e.g. IDT primer quest of primer 3 plus:
• https://www.idtdna.com/Primerquest/Home/Index
• http://primer3plus.com/cgi-bin/dev/primer3plus.cgi
• Then try a PCR using an annealing temp of Tm-2C
• Alternatively, try an annealing temp gradient from Tm-5C to Tm+2C using 1C temp intervals in individual reactions, i.e. reaction #1 = Tm-5C; #2 = Tm-4C... up to Tm+2C. For each of these specific annealing temp set up triplicates to look at how robust your reaction is:
• You might find that at Tm-5C you obtain 1 x specific band and secondary non specific bands; at Tm-2C to Tm abundant specific bands; and at > TM for annealing temp less (o none) of just your specific amplicon

In your case therefore part of the problem might be a mismatch between your primer Tms and your PCR annealing temp; that is your annealing temp is < Tm-2C

To start with there fore you could simply try repeating your reactions but raising your annealing temp by 1 and 2C (respectively): That could lead to consistent specific product amongst all reaction cohorts rather than variation in reaction bands

• For a complete prasy on optimal primer design see attached

 Ali has given some very clear and helpful answers on how to troubleshoot this problem. Maybe what I would add is the simple point:

Primers don't just bind at one site (the site that you were hoping for!).

As Ali explains, primers will bind at any site that has sufficient sequence identity to the primer. The sequence identity required depends on the sequence, annealing temperature, and magnesium concentration. Once the primers have annealed, they will then be extended in the PCR extension step. This means that every PCR reaction will generate quite a variety of products at each step. PCR "success" happens when the desired reaction is sufficiently favourable over all the others that this is the overwhelming product (PCR is exponential, so this should happen). However, if there is another binding site for one primer that will give another product that is as favoured, then you will get two or more bands.

A common control in these cases is to run the PCR with just one of the two primers - that will identify if it is one of your two primers binding at two sites in the template that generates an additional product.

To add to the above:

part of the problem might be a mismatch between your primer Tms and your PCR annealing temp; that is your annealing temp is < Tm-2C

• PCR success is a function of efficiency and specificity
• Efficiency culminating in abundant copies of your PCR product is related to then proportion of targets your primers bind to; especially during the first 5 cycles
• Specificity is ensuring that your primers bind to just one target site
• Specificity can be verified by subjecting your primers to BLAST analysis and ensuring that you only get 100% match for the full length of each primer against the required target
• https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE=BlastSearch
• In addition, to ensure your primers ONLY bind to this target site you need to optimise your annealing temp against your primer melting temp
• Thus, calculate your primer melting temp using freeware such as IDT primer design tool, e.g.
• https://www.idtdna.com/calc/analyzer
• Alternatively, enter your target sequence and allow software to design primers for you, e.g. IDT primer quest of primer 3 plus:
• https://www.idtdna.com/Primerquest/Home/Index
• http://primer3plus.com/cgi-bin/dev/primer3plus.cgi
• Then try a PCR using an annealing temp of Tm-2C
• Alternatively, try an annealing temp gradient from Tm-5C to Tm+2C using 1C temp intervals in individual reactions, i.e. reaction #1 = Tm-5C; #2 = Tm-4C... up to Tm+2C. For each of these specific annealing temp set up triplicates to look at how robust your reaction is:
• You might find that at Tm-5C you obtain 1 x specific band and secondary non specific bands; at Tm-2C to Tm abundant specific bands; and at > TM for annealing temp less (o none) of just your specific amplicon

In your case therefore part of the problem might be a mismatch between your primer Tms and your PCR annealing temp; that is your annealing temp is < Tm-2C

To start with there fore you could simply try repeating your reactions but raising your annealing temp by 1 and 2C (respectively): That could lead to consistent specific product amongst all reaction cohorts rather than variation in reaction bands

• For a complete prasy on optimal primer design see attached

hi reserchares 

from my work ,noticed that some sets of primer give different size of amplicons when the desired region between forward and reverse primer are variable region site.

If the same sample produces amplicons with varying length in each PCR, then the issue is more likely to be within your PCR. However, if it is consistent within one sample but different samples produce different length amplicons, then there is a possibility that allelic polymorphism is present between the different samples. It's still most likely that your PCR needs some optimisation (the advice above covers that), but it's always worth checking that your target doesn't have any known polymorphism that would introduce the different sized amplicons you're seeing.

I just to elaborate on something mentioned in 2 previous answers 

I'm surprised nobody has mentioned splice isoforms yet. If your template is cDNA generated from RNA, and there are several splice isoforms of your mRNA of interest, it's possible to have several PCR products from the same primer pair, these products would differ in exon composition and hence in lenght (usually you would see just 2 bands in this case, but once I observed 5 bands in experiment with splicing inhibitor, there were also pathological isoforms with intron retention and/or exon skipping, and all exons and introns in this template were quite short, 100-300 bp).

Also if you have your target band and pale additional bands it could be false priming (long bands) or primer dimers (short bands

You did not mention if you are amplifying the same sample or a set of samples from different sources. In the latter case you might be looking a variable region of some sort, bit like described by Ekaterina above. 

I would look the Blast-results first as this is effortless and involves no cost or lab labour. As soon as you are sure that there is only one target, check the optimum annealing temperature as described above. If you have unspecific targets, these could randomly be favoured if your starting material (DNA) has variable concentration and/or quality.

• I think isoforms have been omitted from the discussion so far because I take DNA to mean Genomic DNA instead of cDNA. But yes different isoforms if we are talking about RT PCR is likely if PCR conditions appear to be optimal
• That said if you BLAST your primers you should be able to identify such issues as part of a specificity search
• To run the isoform explanation, isoforms tend to be tissue specific so it might be useful as alluded to above to carry our PCR on multiple samples; particularly from multiple tissue types (as a point of principle)
• Before doing this however, do check that PCR conditions are optimised in terms of the annealing temp approximating to Tm-2C of your primers and that primer have been verified for specificity by BLAST; Including if relevant clarifying isoform structures in Ensembl/ Gen Bank (NCBI)