I agree with Rajeshree that the best solution is well designed primers but you can often eliminate primer dimer by using a hot start enzyme. These are inactive at low temperature which is when p-d starts and only activate at 94-96c so if the temperature never drops below the primer annealing temperature you do not get p-d. The other way for only a small number of samples is to make up the pcr mix without enzyme and hold it at 80c. Then add the enzyme,seal the tube and immediately start cycling and again the temperature never drops low enough to produce p-d
I agree with Rajeshree, if your chosen primers have too much complementarity, there is very little you can do about it. Modifying temperature and salt concentrations, will also affect the effectiveness of the PCR reaction. Use a good PCR primer design tool to help you avoid primer dimers, hairpins etc.
Decrease your primer concentration, use a gradient PCR to find the ideal annealing temperature, and use a hot-start enzyme.
Honestly, unless the dimers are interfering with the next step in analysis, sometimes you just don't worry about it. Make SURE you include a water control so you can be sure that what you are seeing is a primer-dimer and not contamination. Good luck!
FIRST, make sure that this specific primer for your target gene to get you target PCR product by using primer design and alignments programs on NCBI..
Then , if you get your PCR product and primer dimer, you should decrease your primer concentration and volume( note that : 10 p.mol of primer is more efficient to get your PCR product)
if you have primer dimer without any PCR product result , check your primer design well and viability.
You can verify that your primers only recognize the desired product, using programs on the design of primers. If they are properly designed, try to decrease their concentration in the PCR or try to make changes in the hybridization temperature of the PCR.