Previous studies on grammatical gender in French individuals with Williams syndrome (WS) have led to conflicting findings and interpretations regarding keys abilities--gender attribution and gender agreement. New production data from a larger WS sample (N=24) showed that gender attribution scores in WS participants exactly mirrored those of controls: all groups overwhelmingly relied on the masculine as the default gender. WS participants' agreement scores were far lower than those of CA-controls though not significantly below those of MA-controls. They also did not improve with age, which might suggest a permanent disability in this area.