After further exploration I have found the following . Using a more explicit query: select distinct aliquot.sample_id, u_protein_sequence as protein, u_dna_sequence as dna from lims_sys.aliquot aliquot, lims_sys.test test, lims_sys.test_user test_user where test.aliquot_id = aliquot.aliquot_id and test.test_id = test_user.test_id and ( (u_protein_sequence = 'F' and u_dna_sequence ='F') or -- Line1 (u_protein_sequence = 'F' and u_dna_sequence ='T') or -- Line2 (u_protein_sequence = 'T' and u_dna_sequence ='F') or -- Line3 (u_protein_sequence is null and u_dna_sequence ='F') or -- Line4 (u_protein_sequence is null and u_dna_sequence ='T') or -- Line5 (u_protein_sequence = 'T' and u_dna_sequence is null) or -- Line6 (u_protein_sequence = 'F' and u_dna_sequence is null) or -- Line7 (u_protein_sequence is null and u_dna_sequence is null) or -- Line8 (u_protein_sequence = 'T' and u_dna_sequence ='T') -- Line9 ) and test.test_template_id = 4 and aliquot.status in ('P','C','V','A') There are only a very few rows in the table where u_dna_sequence is null. The format of the row that contains the null is always protein= T, DNA = null. If line 7 (which corresponds to this pattern) is commented out, then everything runs ok with no errors. If it is left in, then the query will run correctly the first time, but everytime after that, random dna values are set to None in the ZMI zsql test display. Since my DTML that formats the output sets calls None a "no", if something is really a "yes", this is a problem. Any ideas why the ZMI or zope is doing this? Is there known issues with handling null values? Thanks, Mark