The pathophysiological basis of non-syndromic orofacial cleft (NsOFC) is still largely unclear. However, exome sequencing (ES) has led to identify several causative genes, often with reduced penetrance. Among these, the Rho GTPase activating protein 29 (ARHGAP29) has been previously implicated in 7 families with NsOFC. We investigated a cohort of 224 NsOFCs for which no genetic pathogenic variant had been identified by diagnostic testing. We used ES and bioinformatic variant filtering and identified four novel putative pathogenic variants in ARHGAP29 in four families. One was a missense variant leading to the substitution of the first methionine with threonine, two were heterozygous frameshift variants leading to a premature termination codon, and one was a nonsense variant. All variants were predicted to result in loss of function, either through mRNA decay, truncated ARHGAP29, or abnormal N-terminal initiation of translation of ARHGAP29. The truncated ARHGAP29 proteins would lack the important RhoGAP domain. The variants were either absent or rare in the control population databases, and the loss of intolerance score (pLI) of ARHGAP29 is 1.0, suggesting that ARHGAP29 haploinsufficiency is not tolerated. Phenotypes ranged from microform cleft lip (CL) to complete bilateral cleft lip and palate (CLP), with one unaffected mutation carrier. These results extend the mutational spectrum of ARHGAP29 and show that it is an important gene underlying variable NsOFC phenotypes. ARHGAP29 should be included in diagnostic genetic testing for NsOFC, especially familial cases, as it may be mutated in ∼4% of them (4/97 in our cohort) with high penetrance (89%).