I don’t think you can extrapolate from MTT data. I would be far more comfortable doing counts from each sample. 

Thanks for your reply. Dead cells can be stained with trypan blue and counted using hemocytometer. This method is reliable? Any other method is there?

Yes, this method is reliable. You could also use flow cytometry, but it would likely be more time consuming and definitely more expensive. 

Thanks for your reply

Alternatively, you can use a low magnification objective to take microscope images on each well using simple giemsa-staining, or DAPI for fluorescent. Then just use ImageJ to automatically count the nuclei on each picture. If the cell distribution is even on the surface, you can use a higher magnification objective to select a representative FOV. 

You can seed a known number of cells in 96 wells plate to make a curve of number of cells and MTT absorbance. It May be used to estimate the number of cells

Thanks for your reply sir. I have a doubt. I plotted standard growth curve for a cancer cell line (48 and 72 hrs). Attached growth curve with this message sir. For example, If I seed 10000 cells in 96 well plate and incubate for 24 hrs. Then I add chemotherapeutic drug at various concentration and incubate for 48 hrs. Then I did MTT Assay. Now total time = 24+48= 72 hrs. Here which growth curve one has to use 48 or 72 hrs to know no of cells? 

Just for example, Here I am using 72 hrs growth curve, and find absorbance value is 0.3.  with the help of 72 hrs growth curve, I find number of cells were 8000. 

Number of cells killed = 10000(initial amount of cells) - 8000 (final amount of cells)

Number of cells killed = 2000

% of cells killed = 2000/10000 = 20 % 

This calculation is correct? 

I really doubt MTT is the way to go. The MTT reduction will depend on the amount of time the MTT was allowed to reduce, so you would have to ensure that you are using the exact same time each time you do the experiment for the standard curve to work. Stains will work, but they will likely take as much time as a simple TB exclusion assay. You can count one sample per treatment group at time 0, 24 h, 48 h, and 72 h. If you are only looking at the original seeding number, as well as the final number, you are not giving credit to any changes that would otherwise have occurred as the cells doubled in the control group. 

Creating a standard curve with known quantities at the time of readout is a well accepted strategy for many different assays...to measure protein in a Bradford assay, measure cell viability with Cell Titer Glo, determine analyte concentrations with from an ELISA etc. The key is that the standard curve is created at the time of readout to eliminate day to day variability and development timing. I assume this would work for your assay since you probably set up your longest time point first and then run MTT when your shortest time point is complete. The entire experiment and the standard curve really need to be read at the same time.