The aim of this study was to determine whether l-methyl-[11C]-methionine (MET) positron emission tomography (PET) allows the prediction of outcomes in patients with head and neck mucosal malignant melanoma treated with carbon ion radiation therapy (CIRT). This was a retrospective cohort study involving 85 patients who underwent a MET-PET or MET-PET/computed tomography (CT) examination before and after CIRT. MET uptake in the tumour was evaluated semi-quantitatively using the tumour-to-normal tissue ratio (TNR). Local recurrence, metastasis, and outcome predictions were studied in terms of TNR before CIRT (TNRpre), TNR after CIRT (TNRpost), and the TNR change ratio. Kaplan-Meier curves revealed significant differences between patients with higher TNRpre values and those with lower TNRpre values in regard to local recurrence, metastasis, and outcome (log-rank test P<0.0001 for all three). There were also significant differences in metastasis rates and outcomes between patients with higher and lower TNRpost values (log-rank test P=0.0105 and P=0.027, respectively). The Cox proportional hazards model revealed TNRpre to be a factor significantly influencing the risk of local recurrence (hazard ratio (HR) 29.0, P<0.001), risk of metastasis (HR 2.67, P=0.024), and the outcome (HR 6.3, P<0.001). MET-PET or MET-PET/CT is useful for predicting the outcomes of patients with head and neck mucosal malignant melanoma treated with CIRT.