PURPOSE: Translocator protein (TSPO) is a biomarker of neuroinflammation that can be imaged by PET using [(1)(1)C]-(R)PK11195. We sought to characterize the [(1)(1)C]-(R)PK11195 kinetics in gliomas of different histotypes and grades, and to compare two reference tissue input functions (supervised cluster analysis versus cerebellar grey matter) for the estimation of [(1)(1)C]-(R)PK11195 binding in gliomas and surrounding brain structures. METHODS: Twenty-three glioma patients and ten age-matched controls underwent structural MRI and dynamic [(1)(1)C]-(R)PK11195 PET scans. Tissue time-activity curves (TACs) were extracted from tumour regions as well as grey matter (GM) and white matter (WM) of the brains. Parametric maps of binding potential (BPND) were generated with the simplified reference tissue model using the two input functions, and were compared with each other. TSPO expression was assessed in tumour tissue sections by immunohistochemistry. RESULTS: Three types of regional kinetics were observed in individual tumour TACs: GM-like kinetics (n=6, clearance of the tracer similar to that in cerebellar GM), WM-like kinetics (n=8, clearance of the tracer similar to that in cerebral WM) and a form of mixed kinetics (n=9, intermediate rate of clearance). Such kinetic patterns differed between low-grade astrocytomas (WM-like kinetics) and oligodendrogliomas (GM-like and mixed kinetics), but were independent of tumour grade. There was good agreement between parametric maps of BPND derived from the two input functions in all controls and 10 of 23 glioma patients. In 13 of the 23 patients, BPND values derived from the supervised cluster input were systematically smaller than those using the cerebellar input. Immunohistochemistry confirmed that TSPO expression increased with tumour grade. CONCLUSION: The three types of [(1)(1)C]-(R)PK11195 kinetics in gliomas are determined in part by tracer delivery, and indicated that kinetic analysis is a valuable tool in the study of gliomas with the potential for in vivo discrimination between low-grade astrocytomas and oligodendrogliomas. Supervised cluster and cerebellar input functions produced consistent BPND estimates in approximately half of the gliomas investigated, but had a systematic difference in the remainder. The cerebellar input is preferred based on theoretical and practical considerations.