1. The technical replicates.
A technical replicate is when you test the same sample multiple times - it's used to test the variability in the testing protocol itself.
The reason you do technical replicates is to make sure they are almost identical. Triplicate samples and standards are necessary fpr qPCR as there can be pipetting errors when you are adding such a small amount of cDNA that could lead to large differences in your delta cT values. For example, if you get 3 Ct values out of your replicates: 27,3; 27,9; 35. This was pipetted from the same cDNA. So obviously, something wrong happened to the 35 Ct value. You can probably confirm it with the melting curve. And in any case, it is fair (not perfect!) to eliminate your 35 Ct value from your average. Now imagine you did a 2-plicate. Your values are 27 and 35. What is the good one? You average them? No, the only option would be to set this sample aside from analysis.
2. The biological replicates.
A biological replicate is where you perform the same test on multiple samples of the same material / type of cells / tissue. The samples are different, but are expected to be very similar (if not identical) with regard to the test.
Biological replicates are used to test the variability between samples that were selected on the basis of being otherwise identical. According to qPCR analysis model, you will need at least 3 biological replicates. No statistics can be done on less than 3 samples in any case. For example, you want to compare IL6 expression in macrophages stimulated or not with LPS. Then you need at least 3 untreated cultures of macrophages and 3 LPS-treated cultures. But very likely, you will need more for subtler effects where the gene induction is less than 10-fold. Plan 6-10 biological replicates for something that is not very strong.
Souce: NovoPro 2018-02-22