The 1.7% figure is from this article (emphasis mine):
Another study puts the value around 2% (again emphasis mine):
We surveyed the medical literature from 1955 to the present for studies of the frequency of deviation from the ideal male or female. We conclude that this frequency may be as high as 2% of live births. 
The ISNA has a article on how confusing this can be (link) and they don't give out a clear definition, so I won't quote that one.
But in their "10 intersex myths", they state:
In practice, the term “intersex” is used to refer to anybody who was born with anatomy other than what the Powers That Be define as “standard male” or “standard female.”
Note that it only states anatomical differences, while the studies above include other categories.
Other source dispute the broadness of the definitions used by those articles  - from their abstract (ellipsis mine, follow previous link for the full text):
(...) Many reviewers are not aware that this figure includes conditions which most clinicians do not recognize as intersex, such as (...) . If the term intersex is to retain any meaning, the term should be restricted to those conditions in which chromosomal sex is inconsistent with phenotypic sex, or in which the phenotype is not classifiable as either male or female. Applying this more precise definition, the true prevalence of intersex is seen to be about 0.018%, almost 100 times lower than Fausto‐Sterling's estimate of 1.7%.
So the 1.7% is valid if you agree with Fausto-Sterling et alia definition of intersex, that is contested as too broad by other authors.
 - Blackless, Melanie, Anthony Charuvastra, Amanda Derryck, Anne Fausto-Sterling, Karl Lauzanne, and Ellen Lee. 2000. How sexually dimorphic are we?
 - Am. J. Hum. Biol. 12:151–166, 2000. © 2000 Wiley-Liss, Inc. Review and synthesis. American Journal of Human Biology 12:151-166.
 - Sax, Leonard: How common is lntersex? A response to Anne Fausto‐Sterling, The Journal of Sex Research Vol. 39 , Iss. 3, 2002 (paywalled)