Summary: There is evidence to support the figure, but it is misleading.
While the OP dismisses surveys out of hand as untrustworthy, they actually can shed light on the issue, and explain the higher-than-might-be-expected level of 1-in-3 porn viewers being female.
For example, consider this survey of 316 men and 372 women.
Note: This is limited to Denmark, young adults and heterosexuals, and it is only a single study (the first I found; I didn't cherry-pick.) However, it is sufficient to address the question, by revealing how the original statistic was misleading, and not as astonishing as it originally sounds.
While this is a survey and participants may have lied in the answers, we might expect that there would be more cultural pressure for women to:
decline to respond or,
if they did respond, to downplay their porn consumption.
In fact, women were significantly more responsive to the survey request than men, so the first concern can be discarded:
the response rate of the final sample was 65.6% for males (n = 316) and 78.0%
for females (n = 372) (p < .05).
If the second concern was to come about, it would show women's porn-viewing rate as lower-than-actual, which would reinforce our initial expectations. If our expectations are broken despite this bias, we can be confident that our initial expectations of male dominance in porn-watching are incorrect.
Now, one of the original sources of the claim said:
The data, provided to the Herald by Nielsen Net Ratings/NetView, a world leader in internet analysis, reveals 4.3 million Australians viewed pornography or visited a sex-oriented matchmaker site on the internet at least once in the quarter ending in March.
The Danish survey didn't ask about the last 3 months and didn't include sex-oriented match-making web-sites, but they did ask "Have you watched pornography in the last 6 months?" and "last month?" Here are the results (adapted from Table 3).
Men Women
6 months 92.2% (n=309) 60.0% (n=295)
Month 82.5% (n=309) 33.6% (n=295)
Back-of-the-envelope suggests that women therefore account for between 29% and 39% of porn-viewers. The original 1-in-3 claim is right in the ball-park.
Again, if women were more reticent than men to admit to porn viewing, the actual percentage would be higher than this sampled one. If sex-oriented matchmaking sites were included, like they were in the original claim, it would be higher still.
But this figure is misleading, because the number of times that the men viewed the pornography during that 1, 3 or 6 month period is much higher than then women.
From the same table compare the number of respondents who had viewed pornography in the past 24 hours:
Men Women
24 hours 26.2% (n=309) 3.1% (n=295)
Suddenly women only account for 1-in-10 porn-viewers.
This is explained further by another part of the same table, which is limited to only those who indicated they watched some porn, ever:
Frequency of use Men Women
Less than once a month 15.3% (n=281) 51.4% (n=275)
1–2 times per month 17.1% (n=281) 30.3% (n=275)
1–2 times per week 28.8% (n=281) 11.4% (n=275)
3 times per week or more 38.8% (n=281) 6.9% (n=275)
Most women who did view porn used it less than once per month. Most men who did watch porn viewed it more than once per week.
So, from the perspective of a pornography web-site, a particular download of a piece of pornographic material is far more likely to be viewed by a male than a female. (Even more so, as men are more likely to watch pornography alone than women.)
In conclusion: Despite the concerns about surveys, they can be used to explain how the original claimants shared a statistic that was plausible in accuracy, but was presented in a manner that was misleading, in that it over-emphasized the level of women's viewing of pornography.