I'd want to see a large randomized study of the long term effects of melatonin use in adults to establish the long term effects of melatonin use in adults.
How big a study do you need before you'll judge something as safe? You selected two examples of a class of therapies that "appeared safe in the short term, exploited a regular physiological process and had a long track record of use."
From the wikipedia article on the HRT study:
The risk in current users was increased about 1.2 fold; for every 1000 women using HRT, 2.6 developed ovarian cancer over 5 years, compared with 2.2 in those not taking HRT.
The reason huge studies were required to find issues with HRT is because HRT so rarely causes issues. The question you should be asking is: If it is known that a drug is safe in the short term, exploits a regular physiological process, and has a long track record of use, what is the chance that it is harmful in the long term (and to what degree)? The two examples you pointed out are not the entire data set. Your behavior is extremely risk-averse compared to other choices you make daily.
How big a study do you need before you'll judge something as safe?
I don't think it makes sense to ask that question in isolation. When judging whether some risk is worth taking I'd generally look at both the evidence for the potential risks and for the potential benefits. I focused on the potential risks in my original post but the reason I'm not convinced that taking melatonin on a long term basis is justified is that the evidence for the benefits is also weak. If there was extremely strong evidence for the claimed benefits of taking melatonin over the...
Full essay: http://www.gwern.net/Melatonin