I see what you mean, for nested row headers -- which is what my VAR1*VAR2 implied in "PROCTABspeak".  I should have written VAR1, VAR2. Here's the output: And here's the messy dataset when the resulting ODS Excel file is imported:  data are out of register with their nested row headers. I was hasty in writing VAR1*VAR2, I should have written *VAR1, VAR2*n  (...VAR2*colpctn VAR2*rowpctn) ", ie, not nested rows, but a simple single level of row, single level of column crosstab (with corresponding column and row percents if I like): %if &colrow=CRb %then /*NNNN CCCC RRRR*/ TABLE &var1=' ' ALL='TOTAL', (&var2=' ' ALL='Total')*n='Number'*f=comma10. (&var2=' ' ALL='Total')*colpctn='Percent'*f=&Cpctformat (&var2=' ' ALL='Total')*rowpctn='Row percent'*f=&Rpctformat) ; I think this kind of PROC TAB output might export/import more cleanly, but I haven't tried it yet.  ... View more

I have the same issue/need; it now seems to me an idea is to... use PROC TABULATE to generate the exact VAR1 * VAR2 table I want, and then output that table to a dedicated Excel or CSV file, and then PROC IMPORT that Excel or CSV file to a SAS file. This way if I do it once for, say, all numerators (OUT=num) and a second time for suitable denominators (OUT=denom), I can merge the two resulting files by VAR1 and have Ns, denominators, and rates per 100K all in the same record the way I need. ... View more