corrcoef returns matrix of NaNs?

Hello. Any ideas of what would be causing corrcoef to return a matrix of NaNs (with 1 on the bottom right)? Both of my input vectors have large amounts of variance.

Thanks!

[Walter Roberson]

Are there any NaN in your matrix?

[Star Strider]

Based on Conversion of cell containing numbers into a numerical double?, they might not all be numeric. (We do not have the data.)

[Andie Dola]

There are no NaNs in either of the inputs. I am using a separate data set from my previous question.

[Walter Roberson]

Are there any inf ? What does cond() of your two matrices report?

[the cyclist]

Can you attach a *.mat file with your vectors?

Also, one way to debug this would be to trying running corrcoef first on one half your observations. If you do, cut it in half again. (If not, run it on the other half.) This way, you should be able to discover which of your observations are causing the problem.