This study aimed to determine the relationship between postural asymmetry and falls in Parkinson's disease (PD). In total, 28 patients with PD were included. Postural control was analyzed in bipedal, tandem, and unipedal standing. Center of pressure (CoP) parameters were calculated for both limbs, and asymmetry was assessed using the asymmetry index. Logistic regression was used to predict/classify fallers through postural asymmetry. The Spearman correlation was performed to relate asymmetry and falls number. Poisson regression models were created to predict the number of falls in each condition. The results demonstrated that asymmetry can classify 75% of fallers and nonfallers. Asymmetry in anteroposterior-mean velocity of CoP in unipedal standing was related to the number of falls. Poisson regression showed that anteroposterior-mean velocity of CoP predicts falls in PD, indicating that increased asymmetry results in a greater number of falls. Anteroposterior-mean velocity of CoP seems to be a sensitive parameter to detect falls in PD, mainly during a postural challenging task.