Correct Evo7. Don't see much of x**3 or references to the Fortran version of Numerical Recipes these days! . .. .I don't think Excel is the right tool for the job but if it is a requirement that the wrong tool be used, then one of Excel's curve-fitting algorithms can be used (linear, logarithmic, polynomial, power, exponential). On the Options tab select "Display equation on chart". This will produce an equation of the form y = f(x). It will be smooth and therefore have a tangent line. The slope at (c,f(c)) can then be calculated by (f(c+h/2)-f(c-h/2))/h where h is a small number. How small? Depends on how many decimal places Excel uses. If single precision, I'd try h = E-04; if double, then h = E-06 will work fine. Because of "loss of significance" too small a value will give inaccurate results.. .. .One caveat - a curve might fit data very well without its slope doing the same. For example, y = x + sin((10**20)*x)/(10**14) is a good fit of y = x but its slope is not close to 1 for most x.. .. .As Samuel Johnson said of graphing with Excel: "It is not done well; but you are surprised to find it done at all." Maple is the thing to use.