To investigate the relationship between temporalis muscle thickness (TMT) at baseline as a surrogate marker for sarcopenia and long-term motor outcomes in patients with Parkinson's disease (PD). We enrolled 249 patients with drug-naïve early-stage PD (119 males and 130 females, follow-up > 3 years). Baseline TMT of each patient was measured on the axial plane of T1-weighted images. The association between baseline TMT and long-term motor outcomes in PD was assessed using Cox regression models for levodopa-induced dyskinesia, wearing-off, and freezing of gait and a linear mixed model for the longitudinal increases in levodopa-equivalent dose per body weight over time. Statistical analyses were performed separately for sex if an interaction effect between TMT and sex was assumed. TMT differed substantially between the sexes, and male PD patients had higher TMT (6.69 ± 1.39 mm) than female PD patients (5.64 ± 1.34 mm, p < .001). Cox regression models demonstrated that baseline TMT was not associated with the risk of developing levodopa-induced dyskinesia, wearing-off, or freezing of gait during the follow-up period. The linear mixed model was applied separately for sex and demonstrated that higher TMT at baseline was associated with slower increases in levodopa-equivalent dose per body weight in male PD patients, but not in female PD patients. This study demonstrated that baseline TMT could be an indicator of the longitudinal requirement for dopaminergic medications in male patients with PD, suggesting that sarcopenia may have a detrimental effect on disease progression in PD in a sex-specific manner.