Excel is not a program designed for statistical analysis, although it comes in handy when all you want to do is some quick and dirty analyses. However, you should be cautious about the results, as some can be simply wrong. Here is one problem from my experience with Office 2008 for Mac.

When doing exponential regression, Excel calculates the R-squared value based on transformed data. For example, for a exponential curve y = a * ebx, Excel computes R² using ln(y) = ln(a) + bx. Needless to say, this can cause some unwanted surprises. With some quick google searches, you can easily find more problems.

So there, you have been warned.