Hi there I have prepared my standard curves for qPCR but Im not quite sure what you are asking. I have also used a qubit for quantifying both DNA and RNA but not in the context of cDNA. However, I think the answer is 'yes you can' because the usual problem when measuring synthesised cDNA (unless column purified post RT) is contaminating dNTPS which qubit fluorimetry does not detect. Here is a general background answer providing context but the simple answer is qubit quantitation unlike 260nM absorbcance is correct. BUT YES IN ESSENCE WHAT YOUR DOING IS CORRECT BUT DOES SUGGEST DIFFERENCES IN C DNA SYNTHESIS EFFICICNCIES BETWEEN YOUR SAMPLES IMPLYING DIFFERENT RNA QUALITIES. See explanation and context below:

Generally speaking there are 2 ways of ensuring that samples are normalised for gene expression for qPCR: Either you transcribe from a known and constant amount of total or mRNA - say 10ng (mRNA) or 200ng (total RNA) and then put a set amount into your qPCR reaction (say 1ul) including samples for standard curves based on serial dilutions. Quantifying at the start is normally done with RNA that is reverse transcribed in tubes in vitro and not subsequently purified. In this instance you cannot quantify the cDNA at the end as a big part of your signal would be unincorporated dNTPS if you base amounts on readings at 260nM.The disadvantage of this ouset quantitation is that if the efficiency of RT is different for different samples - say based on purity of starting RNA - then you will end with different amounts of cDNA sample to sample 

Alternatively, you can perform RT on a column and purify as part of the column elution step  which will also lead to purified cDNA devoid of contaminating dNTPS. In this instance you can measure your cDNA and correct the dilutions accordingly, using either qubit fluorimetry or spectrophotometry

To finish my answer I would say that qubit quantitation of cDNA and subsequent correction of dilutions for your RNA serial dilutions is probably the best thing you as it measures the final cDNA (without detecting contaminating dNTPS unlike 260nm spec) and thus takes into account differences sample to sample in cDNA synthesis. For RNA samples with 260/280 ratios of 1.8-2.1 and 260/230 ratios > 1.0 then putting a standard amount of RNA as decsribed into your RT reaction and then diluting or using a standard amount of RNA (typically 1ul from a 20ul or 50ul total reaction) will usually work consistently well. The fact that it doesnt for you and you are having to make qubit corrections - which to re iterate is completely OK - suggests differences in RNA quality; ergo potential differences in 260/280 ration of your starting RNA and especially and potentially differences in the 260/230nm ratio; specifically some samples with 260/230 ratios of > 1.0 and upto about 1.8 (ideal samples) and some samples with ratios < 1.0 (contaminated with GTC and likely to reverse transcribe poorly and thus rquire more of the final cDNA in the final serial dilution based on qubit readings). Specifically 230 nM measures the absorbance of gianidium iso thi0 ocyanate in your lysis buffer whic put into the lysis buffer to denature RNAses and into the first RNA purification wash buffer to denature proteins. If this bleeds through into your final eluted RNA then it can potentially inactivate reverse transcriptase and thus lead to poor RT reactions. In my experience samples with 260/230 ratios of < 1.0 implyina guanidium ITC contamination perform poorly in RT reactions and could lead to the scenario you are describing.

Thus if some of your samples have 260/230 ratios < 1.0 then the only legitimate way to perform an accurate standard curve would be to use the qubit (to account for differences in RT efficiency). Irrespective though it is generally a good thing (although not strictlly necessary unless dealing with the above) so go ahead !!    

Finally, if the source of differnces between standard dilution and qubit correction of your samples is caused by GITC cxontamination of your staryting RNA then there are modified protocols - which I can provide for improving removal of GITC and thus increasing your 260/230 ratio

I think Ive said enough and hopefully provided backgound explanation but bottomn line is yes go  check for this sort of RNA  quality variation in your own samples)

If you have any further questions do not hesitate to contact me

http://www.ncbi.nlm.nih.gov/pubmed/23649952

It seems that your dilutions are not what you wanted them to be. So there is obviousely a problem with your way of diluting the standard. I would not trust in any way "corrected" results when I have doubts that the researcher can pipet properly (this is a rather harsh formulation - I know that there can be many difficulties other then "clumbsy pipetting", but whatever the reason is: somwhere in your assay is a problem, and introducing other "corrections" rather than solving (or smatrly circumventing) the problem is another problem...)

thank you for your interesting answers,  you are right Jochen, the linearity of my dilutions and the limit of detection of my instrument seems to be the problem, since is the last dilution the one tha is giving me the hard time

I agree that you should resolve the issue at the source rather than going around it. If your are having poor reverse transcription for any reason, you will not be getting accurate qPCR results since the mRNA of interest may or may not have made it into cDNA form for detection. In my qPCR protocol, we select for mRNA (using oligo dT) and reverse transcribe 5 μg of total RNA in 20 μl reaction, which is the maximum recommended in SuperScriptIII kit, then use 1 μl of 20x dilution of cDNA  for detection purposes. You can try standard curve around that dilution and see if it works. My quantification for the total RNA is based on NanoDrop.