nanpos = find(isnan(YourData(:,3))]);
nanruns = diff([1 nanpos size(Yourdata,1)]);
blocks = cell2mat(YourData, nanruns, size(YourData,2));
blocks( cellfun(@isempty, blocks) ) = [];
Each of the blocks in the cell array will now start with a NaN entry, except in the case where the first entry in the array did not have a NaN, in which case the first block will include all the entries before the first NaN. Only the first block has that possibility.
If the last entry in the array is a NaN entry then the last block will be that entry by itself. I did it this way, instead of deleting the NaN, in case the NaN entry has some useful information you want to extract.
