A good review is: KOTCHEN T A 2011, Hypertension 58; 522.

I personally think that the cuff measurement of BP is antiquated and should be replaced by a technique that takes into account an individual's perfusion of essential organs & cardiac function, which means one size does not fit all. But inherent conservatism of clinical medicine, insurance companies, & drug manufacturers will probably prevail. Last I heard we were all trying to be 115 systolic.

So the short answer is: nobody discovered it..

Thanks Usman and Alan Coulson. I have strong doubt that the 'normal values' for any physiologic parameters (BP, cholesterol, blood glucose etc) must be researched thoroughly again with sound correlations with physical efficiency, body composition (fat%) and somatotype. It is also highly possible to find people with 120/80 mmHg of BP but very poor physical efficiency with unfavorable body composition and disturbed homeostasis.

The question I'd ask is: is there a threshold for blood pressure below which the risk for adverse outcomes, be in mortality, MI or whatever, no longer changes or actually increases? 

Here's data from the NHANES dataset (if you're unfamiliar, just health screening for several thousand random americans that the US does every 10 or so years). Looking at the graphs for the under 65 years old crowd, I'd say that all cause mortality continously goes down the lower your systolic BP and hits a minimum around 80 diastolic. So you might argue that 100/80 is the normal BP.

The problem is that this is just correlation. What we really care about is targets for treatment and for that you need a randomized trial. Take a peek at the SPRINT trial. The 120/80 target did better than the 140/90 target, so I'd take this as the more definitive evidence that 120/80 is the "normal", or more correctly, "optimal" blood pressure.

Article Systolic Blood Pressure, Diastolic Blood Pressure, and Pulse...

Article A Randomized Trial of Intensive versus Standard Blood-Pressure Control

The way I understand the question: it is one of Medical History. None of the replies have addressed that interesting aspect. In the famous 1905 publication by Korotkov there is no number affixed to the first ausculatory measurements. I suppose that the first 'ideal, normal values' got settled around 1930, when the relation between kidney disease and hypertension became clear. But I have not found literature, yet, where 120/80 became the gold standard for blood pressure.

In my experience, I have come across individuals with BP values above 120/80 (sometimes even 160/100 or even more in resting state) are taking part in intense strength training workouts without any adverse effects. On the other hand, individuals with BP values like 110/70, 120/80 have very poor exercise capacity too. How 120/80  got generalized as normal BP needs broader explanation with evidence of what type of experiments conducted on whom and what they had in common other than 120/80.

http://trialsjournal.biomedcentral.com/articles/10.1186/1468-6708-6-5

Dr. Rajkumar brings up another excellent point; these measurements are made with the patient at rest. As such they have little bearing on the BP during normal activities such as being at work, recreational activities and so on.