The standard approach is to do a kill curve for your cells to determine the puromycin dose necessary to kill your cell line of interest so that any cells that grow in that concentration of your selection agent must be expressing the puromycin resistance conferred by you transfection. The concentration should not be so high as to prevent your resistant cells from growing.

Debanjan, 

You are essentially asking others to guess what level of puromycin N-acetyl transferase expression your specific expression system in the strains of cells you mentioned under your experimental conditions should you expect. Such guesses for the obvious reasons cannot be accurate and as already pointed out by the above two respondents, one must determine the effective range empirically.

The common range is about 1-10µg/ml. But I have had used as high as 20-25µg/ml in some cases. Fortunately puromycin is quite quick in doing its its job so 2-3 days is all you need.

Thanks to Prof Lange, Priyakshi and Dr. Alam for  your opinions. Yes I know about the kill curve and I have already set up an experiment for that to determine the effective range. Although 5 µg/ml looks an effective dose, but the cells selected from this selection are growing very slow. Is this common? 

You may need to set up clones from your selected cells to find a population that has nearer normal growth rates. Note that the puromycin concentration that you use for selection should be considerably higher than that used for maintenance (to prevent back mutations competing with your selected cells). You may wish to grow your selected cells without puromycin to test for normal growth and reselect periodically if they don't grow well under maintenance conditions. You may also need to check the puromycin resistance gene in your construct to see how strong its expression is.