positive controls are any sample that has amplified before and are useful  in checking that the pcr is working so that you can say a negaive result is negative and not just a failed reaction but negative controls containing no DNA are absolutely essential in pcr. They indicate whether you have proper amplification. The powee of pcr is enormous...every 10 cycles about 1000 times more product is generated so only a few molecules of previously amplified product in a reagent will amplify better than intact dna as the contamination is already short and melts well so amplifies well. If your negative control shows bands of the correct size then you can have no confidence that you are amplifying the correct dna sample or just contamination from a previously amplified sample and working with results from the wrong samples is a very bad thing

A positive control depends on the gene you want to amplify. For example, if you are amplifying a 16S gene of E. coli, a positive control could be another already amplified 16S from a similar bacterium. So, in a gel you'd get a clue if the amplified band correspond to your gene.

Controls in PCR means you can find out what you wants to amplify. If you run both the controls, negative and positive, the you confidently say all about your PCR results. 

Lets us consider I wants to work on viral DNA e.g. human parovirus B19. 

I will prepare total mastermix of for 100 ul total reaction and divide in five equal aliquotes. My mastermix contains everything except genomic DNA. 

I will prepare my experiment as follows:

1. aliquote one: negative control (I will add only nuclease free water)

2. aliquote two to four: my sample no 1, 2 and 3

3. aliquote five : positive control (here I will add NS1 region cloned in plasmid from e.g. Bovine parvovirus, so that my primer region is conserved and will amplify human parvovirus as well with the same set of primers, also as Bovine virus NS1 region between the forward and reverse primer is somewhat different, I may expect different size of amplicon than that of human virus). 

when you do gel analysis, you should not find any amplified product even a traces of that in negative control, that will suggests that your reagents are not contaminated with any positive control or positive sample.  Your positive control should always amplify that suggests that all your conditions and reagents and machines are working fine. Now you can correctly interpret, in aliquote 2, 3 and 4; whether the B19 virus is present or not by simple finding or not finding amplified product in respective wells. always remember to add positive controls at the last in all the steps.  

For bacterial positive control its very easy and explained by Daniel. I just wants to add, this 16S r RNA from bacteria or ITS2 from fungi or plant can act as positive control for amplification of any single copy genes such as rpoB in bacteria. 

for human you can use beta globin as positive control for PCR and GAPDH for RT PCR, 

So depending on what you wants to do, you need to think on positive control. 

positive controls are any sample that has amplified before and are useful  in checking that the pcr is working so that you can say a negaive result is negative and not just a failed reaction but negative controls containing no DNA are absolutely essential in pcr. They indicate whether you have proper amplification. The powee of pcr is enormous...every 10 cycles about 1000 times more product is generated so only a few molecules of previously amplified product in a reagent will amplify better than intact dna as the contamination is already short and melts well so amplifies well. If your negative control shows bands of the correct size then you can have no confidence that you are amplifying the correct dna sample or just contamination from a previously amplified sample and working with results from the wrong samples is a very bad thing

THANKS a lot Devendra purushottam Lingojwar for answering and good guideline and also thanks to Paul and Daniel