If we limit ourselves to the relatively safe U.S. (no cholera, for example) Giardia lamblia is hardly the only water-borne pathogen to concern us. Cryptosporidium parvum ("crypto"), Campylobacter, and E. coli are also rather significant risks.
According to the Gov., for 1598-2005 there were 798 shark attacks in the waters of the United States, resulting in 39 fatalities -- none, I might add, where I go backpacking.
In 1993, 400,000 people were infected by crypto in Milwaukee. 100 died. Those in good health rarely died. They did suffer an average if twelve (12) days of watery diarrhea, peaking at nineteen (19) watery bowel movements per day and a typical loss of weight of ten (10) pounds.
In an incident in the past ten years, 2300 got sick in Walkerton, Canada, from water-borne pathogens. Seven died.
So water-bourne pathogens are a far more significant risk than shark attacks - especially on land.
The CDC calls water-bourne pathogens "a major cause of morbidity and mortality."
Crypto, like Giardia, is spread by a variety of mammels. Bambi (or Billy) drinks from an infected stream, takes a dump in or near the stream over the ridge, and the second stream is a risk. (Hence the rules in the parks and forests mandating [Pious hope!] hundreds of feet between streams and places where one takes a "dump.")
Estimates for distribution of Giardia that I found on the internet range from "most" to 16 - 17.8% of total surface water sources (Distribution of Cryptosporidium and Giardia in Surface Waters, Abrams, Lechevallier, and Abbaszadegan, Abstracts of the General Meeting of the Society of Microbiology, 1999).
In a survival situation, you are already nearer the edge than the norm. Increasing the risks for an adverse result by drinking untreated surface water seems ill-advised. Twelve days of watery diarrhea, peaking at 19 bm's a day might tip the margin to non-survival.
But suit yourself.
Ed: If you are dying of thirst, you do what you have to do. But as noted by SkunkWerX, because a relatively slight weight penalty solves the problem, why not have that gear? And if time allows, boiling or even slow sand filtration is quite effective. We ought to reason to the correct solution from the facts.
NOTE: Neither iodine or bleach are a safe preventive for Crypto.