Calculating the Pairwise covariance: with NaN's

I am trying to calculate the pairwise covariance.

x=rand(100,36);
TheNaNC=randi([1 100],1,10);
TheNaNV=randi([1 36],1,10);
x(TheNaNC,TheNaNV)=NaN;
c2 = cov(x,'partialrows');

I get the following error message: 'Error using cov (line 64) The number of elements in x and y must match.'

Also tried:

c3 = cov(x,'omitrows');

'Error using cov (line 64) The number of elements in x and y must match.'

What am I doing wrong with this?

I am trying to calculate the covariance the why R calculate it but using MATLAB

[Peter Mills]

I think the answer is to use:

Y = nancov(x,'pairwise')

[Peter Mills]

nancov gives strange results

[Torsten]

The random integers should be drawn between 1 and 100 (1 and 36), not 0 and 100 (0 and 36).

Best wishes

Torsten.

[Peter Mills]

Ok have changed this. My problem is with calculate the covariance with NaN's in the data without throwing away data