You can use the tables in two ways. The usual way is to take a Z value you've calculated and then look up the number (between 0 and 1) corresponding to the Z value. The area under the normal curve is standardised as 1, which of course is 100%. When you look up Z=2.17 you see the number 0.9850 meaning that 98.5% of the area is to the left of Z=2.17. If you look up Z=-2.17 you get 0.0150 or 1.5%. The mean cuts the normal curve into two equal halves. On the right of the mean we have all the positive Z values while on the left we have all the negative Z values. Because of symmetry 98.5% of the area on the left takes you up to Z=-2.17. Outside this range we have 1.5% (Z=-2.17) and 1.5% (Z=2.17, 100-98.5%), making the 3%=100-97 asked for in the question.
The other way to use the table is to look in the body of the table, as I did, and work back to the Z value. If we start with a different percentage we will arrive at a different Z value. All the Z value really is is a measure of how many standard deviations we are away from the mean. So Z=2.17 corresponds only to the number 0.9850. Take a look at the table (also available online) and you'll see what I mean.
I think you probably knew most of what I just mentioned, but I hope I've included the answer to your comment. And I have to say I'm no expert on statistics, so do go for a second opinion on my answer, won't you?