Levodopa effectively improves motor symptoms of Parkinson’s disease. However, the beneficial effects of levodopa often erode over time with the emergence of response fluctuations. Although these response changes have been recognized from the early levodopa era, their mechanisms remain poorly understood. We investigated the role of dopamine (DA) terminal loss in the development of motor fluctuations by employing PET with [¹¹C](±)dihydrotetrabenazine ([¹¹C]DTBZ) as an in vivo marker for DA nerve terminals. Levodopa response was characterized by analysing the time–response curve to a single dose of levodopa with a finger‐tapping test. PET scans were performed in 11 patients with asymmetric Parkinson’s disease (age: 61.12 ± 7.97 years; duration of Parkinson’s disease: 10.55 ± 4.53 years; mean ± SD). Each patient performed finger‐tapping tests for up to 5 h after taking a therapeutic dose of levodopa. Results showed significantly lower [¹¹C]DTBZ binding potential (BP; B_max/K_d) and baseline tapping rates on the more affected putamen and corresponding body side, respectively, than on the other (P = 0.003 for the former, P = 0.013 for the latter). Among the variables describing the time–response curve, the duration and early decay time were significantly shorter on the more affected side (P = 0.051 and P = 0.021, respectively). Latency to the onset and latency to 50% E_max (the magnitude of the levodopa response) were significantly longer on the more‐affected side (P = 0.013 and P = 0.004, respectively). E_max was not significantly different between the two sides. The asymmetry (difference from the more affected to less affected side) of [¹¹C]DTBZ BP in the putamen showed a highly significant correlation with the corresponding asymmetry of the estimated EC₅₀ (levodopa concentration producing 50% of the maximal response; P = 0.022; r = –0.727), a marginally significant correlation with that of latency to the onset (P = 0.065; r = –0.583) and no significant correlation with that of the magnitude, duration or early decay time. This pattern of changes in levodopa response from the less affected to more affected side was similar to that from stable to fluctuating responders except for the latency to onset. These findings suggest a pathogenetic role for DA terminal loss in the development of motor fluctuations. However, the absence of a significant correlation between the early decay of levodopa response and DA terminal density suggests that DA terminal loss alone cannot account for the development of motor fluctuations. Therefore, our study suggests that both levodopa treatment and DA terminal loss contribute to the pathogenesis of motor fluctuations.