Sequencing of human genome samples has unearthed genetic variants for which functional testing is necessary to validate their clinical significance. We used the Drosophila system to analyze a variant of unknown significance in the human congenital heart disease gene NKX2.5 (also known as NKX2-5). We generated an R321N allele of the NKX2.5 ortholog tinman (tin) to model a human K158N variant and tested its function in vitro and in vivo. The R321N Tin isoform bound poorly to DNA in vitro and was deficient in activating a Tin-dependent enhancer in tissue culture. Mutant Tin also showed a significantly reduced interaction with a Drosophila T-box cardiac factor named Dorsocross1. We generated a tinR321N allele using CRISPR/Cas9, for which homozygotes were viable and had normal heart specification, but showed defects in the differentiation of the adult heart that were exacerbated by further loss of tin function. We propose that the human K158N variant is pathogenic through causing a deficiency in DNA binding and a reduced ability to interact with a cardiac co-factor, and that cardiac defects might arise later in development or adult life.