Darlot and colleagues recently reported that near-infrared light is neuroprotective in a monkey model of Parkinson disease (PD). 1 I believe that the authors have adopted an inconclusive design and overlooked several parameters that could have a major effect on their analysis and conclusions. In their design, 2 subsets of animals received different regimens of methylphenyltetrahydropyridine (MPTP) administration on the grounds that these had been used in previous studies. In the cited publications, however, neither of these regimens was used. The data from the 2 subsets were eventually pooled (both the MPTP only group and the animals exposed to near-infrared light). However, within the near-infrared light exposed group, 2 groups were identified post hoc (NIr1 and NIr2) on the basis of their clinical scores. The above constitute serious experimental design flaws. Despite this striking self-contradiction and experimental flaw, the behavioral (Fig 4) and anatomopathological (Fig 6) standard deviations of the so-constituted groups (but obviously not the means) are extremely low. Even more concerning is the relationship between the level of nigrostriatal degeneration and symptom severity. The required extent of degeneration for symptom onset is about 25% of Nissl-stained cells, and 44% of tyrosine hydroxylase (TH)-positive neurons, whereas striatal markers should be decreased by 80%. 2 For very severe, late PD-like symptoms, the required level of neurodegeneration is 65% of Nissl-stained cells, 85% of TH-positive neurons, and >95% of dopamine marker loss in the striatum. 2 The present MPTP group, however, shows a loss of 50% of THpositive neurons, 30% of Nissl-stained cells, and 90% of TH terminals. Such an extent of lesion is inconsistent with the very severe, late PD-like symptoms described by the authors. Quite puzzling, also, is the observation that the MPTP-NIr1 and MPTP-NIr2 groups display extents of nigrostriatal denervation that are slightly below the threshold for symptom appearance. None of the animals should thus show PD-like symptoms.The preclinical testing of a putative neuroprotective strategy must, among other criteria, demonstrate that the tested intervention can protect the neurons that remain once PD symptoms have appeared, that is, once dopamine depletion has reached 70 to 80%. Such a clinically relevant design would aim at mimicking the clinical situation 3 and grounding the translational significance of the findings. In addition, a preclinical design should be statistically valid, an obvious statement for any preclinical work but particularly when dealing with nonhuman primates. Clinical ratings being by nature noncontinuous, the scores should be analyzed with nonparametric tests as per guidelines, 4 which is not the case in the present study. A standard t test is even used for comparisons of mean scores between the 2 groups. Finally, the authors apparently used a 1-way analysis of variance (ANOVA; providing neither the F values nor the degrees of freedom) when a 2-way ANOVA should have been used,...