I think the same, 40 cycles are enough, however I've heard about people doing 50...

Each time when I set up the new reaction I do electrophoresis + sequencing. During the measurements I check the melt curve and rarely, if the peak aren't nice I run the samples on the gel.

Thanks.

I agree with 40

We often use 40 but in fact I would consider reliable what is within the range of your standards