Nonspecific binding (NSB) is a key parameter in optimizing PET imaging tracers. We compared the ability to predict NSB of three available methods: LIMBA, rat fu,brain, and CHI(IAM). Even though NSB is often associated with lipophilicity, we observed that logD does not correlate with any of these assays, clearly indicating that lipophilicity, while influencing NSB, is insufficient to predict it. A cross‐comparison of the methods showed that all three correlate and are useful predictors of NSB. The three assays, however, rank the molecules slightly differently, illustrating the challenge of comparing molecules within a narrow chemical space. We also noted that CHI(IAM) values more effectively predict VNS, a measure of in vivo NSB in the human brain. CHI(IAM) measurements might be a closer model of the actual physicochemical interaction between PET tracer candidates and cell membranes, and seems to be the method of choice for the optimization of in vivo NSB.