How much is enough current for a LED?

I usually drive LEDs at 20mA to get them nice and bright, but that's when I don't have power/current limitations. Now I am designing a board with only a 3.3v rail and it will have about 10 LEDs (though perhaps not all on at the same time). I don't want to draw 200mA just for LEDs.... so my question is, what can be the current rating which is much lower but will still make the LEDs glow nicely. Thanks.

Not all LEDs are made the same. Some are more efficient than others. One batch that I bought 20 years ago are barely usable as power indicators at 5mA while some other batches are almost too bright at that current for the same purpose. Many are bright enough at 1 or 2mA.

In any case, "glow nicely" is too vague a term as it depends on the application and viewing conditions. The most important factor here is what you consider to be bright enough for your application. Light up one or two sample LEDs at different currents and decide which level is good enough.

for indication LED through hole mamimum is 5ma....

Try driving one by one in a loop at speed greater than 100Hz. Average currrent will be for sngle LED. For example drawing power to led from outputs of decade counter like CD4017, driven by a 1Khz clock input.

I agree with post #4, to experiment with strobing the led's.

Additionally you can step up current per pulse. Research has shown the led's have a greater perceived brightness this way.

In other words you can apply 40 mA at a 50% duty cycle without reducing the led's useful life (theoretically). 80 mA at a 25% duty cycle. Etc.

Of course you can only carry this so far and then you have barbequed led's.

When I suggested running a couple of sample LEDs at different current level, what I had in mind was something simpler than strobing or using a counter. A DC source and a few different resistor values would provide a good indication in less than a minute. For example, with a 5V supply and red LEDs, a 1k resistor will supply 3mA, 330 ohms will push 9mA, and so on.

I have a simple gadget that I made some a long time ago that's well suited for tests like this. I'd accumulated a mess of zener diodes over the years and the markings on many of them were illegible. So I built a basic constant current supply that can be varied from 0.5 to 15mA. It became one of the most useful simple gadgets I ever designed and I've been using it ever since for many applications other than the original purpose.