Welcome to nanodrop... well, one thing you can do is to use little more concentrated nucleic acid sample. Most such variations result from the use of too diluted sample. Contrary to popular belief, nanodrop is still a spectrophotometer and follows the rules and laws that are true for any spectrophotometer. 

You can enhanced accuracy while doing following things: Get your Nanodrop calibrated again, take more concentrated solution for reading and mix well your solution before taking reading every time.

There are many reasons to it. You can read the article in the given link. I think it'll help!

http://bitesizebio.com/26691/arent-nanodrop-results-reproducible/

I suggest to get your Nanodrop Calibrated. Normally I take 1:10 dilution of sample, mix it well and take the reading. And have always got constant reading.

 Thanks all for your kind and quick response. I would like to also know that should I need to make the calibration every time for each sample. If so it's quite time consuming work, I think. Anyway thanks everybody

DNA and RNA quantification: fast and simple with

PicoGreen® dsDNA and RiboGreen® RNA quantification reagents

Fluorescence intensity on Infinite™ F200 and Infinite M200

http://neotec.co.il/wp-content/uploads/2014/03/TN_Inf200_Quant-iT-PicoGreen_Infinifte-F_M-200_395180_V1.0.pdf

In addition to measuring samples at a concentration that gives A260nm readings of greater than 0.05 (for DNA samples) and above 0.0625 (for RNA samples), use of the appropriate blank is important:

After allowing the NanoDrop to measure water for its own zeroing purposes, it then usually asks for a "blank" - at which time, many investigators again use water - but, in most cases, one should not use water for this.  The blank should be whatever 'buffer' your DNA or RNA sample is actually in. To create an appropriate blank, one needs to put a water sample all the way through the extraction process one uses for DNA or RNA extraction, then, the resulting 'extracted water' sample serves as the correct blank.

Ideally, one would extract a sample matrix identical to the biological sample one is working with, but which contains no nucleic acid.  This material is hard, if not impossible to come by, but it would indeed represent the best thing to co-extract in order to create the appropriate 'blank sample' - even better than co-extracted water.

MD SAIFUL ISLAM above - has the best idea for these kind of measurements.  But, they require specialized equipment and the ability to handle the associated reagents carefully, correctly, precisely, accurately. In addition, when comparing these fluorescent methods of measuring double-stranded or single-stranded nucleic acids in solution, I find it highly troubling that Qubit and NanoDrop results often poorly agree. [Meaning that:  if one believes the NanoDrop or Spec readings without ever knowing what a fluorescence-based reading might have shown, then many unknowingly ill-rendered NanoDrop and Spec readings are actually useless to unknown/varying degrees].

It is important to realize that the NanoDrop (or any spectrophotometer) 'sees' all nucleic acids and any co-existing NTPs and dNTPs at the same time - it does not discriminate between them (RNA, DNA, cDNA), and free NTPs and dNTPs absorb even more strongly at 260nm.

Using the appropriate blank helps - but not absolutely.  If the RNA sample is contaminated with DNA, or an extracted DNA sample is contaminated with RNA, or a cDNA sample (from an RT rxn) still contains left-over dNTPs and undegraded and degraded RNA (and its associated NTPs), NanoDrop and Spec readings are bound to give vaying degrees of dubious results in many situations. 

Thus, in my case, when I extract RNA by TRIzol method, I am extremely careful to avoid the white protein/gDNA interface at that step in the protocol. Secondly, after isolating the total RNA, DNAse-treating it, adding RNAseOUT, and diluting a portion of it to 1:50 for Spec measurement, I have also co-prepared a sample of water that has also been taken through the entire process as well - through TRIzol, DNase treatment, RNAseOUT addition, and 1:50 dilution - and that is used as the final blank for the instrument.  My purity (260nm/280nm) ratios are consistently above 2.0 for all of my extracted RNA samples using this approach. 

Curiously, even though I am fairly confident that my RNA samples are already DNA-free even before DNAse treatment, I do the DNase treatment step anyway as a precaution just in case, even given the careful avoidance of the gDNA/protein interface layer during the TRIzol RNA-extraction protocol. Left-over components of the DNase-treatment step and RNAseOUT addition appear to be qPCR-friendly afterwards - especially within the optimal working range I use for the qPCR by employing the PREXCEL-Q Method.

Perhaps there are some helpful ideas in what I wrote here -- it seems to work for my particular RNA sample isolates anyway...

Dear Jack M Gallup,

Thanks a lot for your answer and and comparative analysis. Can you please let me know what will be the approximate value for blank (water, buffer, 260nm/280nm and 260/230 nm) because I also found different value for blank for different measurements. One of my friend told me that from we can consider 0 to plus/minus 5, is it ok?

Best regards

Ashraf