Are the Y values something measured that are supposed to be in a straight line but might vary due to measurement errors?
If so, calculating the least-squares-fit line is your best bet. Here's a good link that describes the algorithm and gives you the vb code:
http://www.vb-helper.com/howto_linear_least_squares.html[
^]
The algorithm in that link gives you the equation of the best fit line in terms of a slope (m) and intercept (b) such that:
x = y * m + b
So if you have that, you can calculate y for any given x.
If on the other hand the data isn't expected to be a line, but you just want to fill in the missing y's by linearly interpolating, then Walt Fair, Jr.'s solution works great.