Background Nor-BNI, GNTI and JDTic induce κ opioid antagonism that is delayed by hours and can persist for months. Other effects are transient. It has been proposed that these drugs may be slowly absorbed or distributed, and may dissolve in cell membranes, thus slowing elimination and prolonging their effects. Recent evidence suggests, instead, that they induce prolonged desensitization of the κ opioid receptor. Methods To evaluate these hypotheses, we measured relevant physicochemical properties of nor-BNI, GNTI and JDTic, and the timecourse of brain and plasma concentrations in mice after intraperitoneal administration (using LC-MS-MS). Results In each case, plasma levels were maximal within 30 min and declined by >80% within four hours, correlating well with previously reported transient effects. A strong negative correlation was observed between plasma levels and the delayed, prolonged timecourse of κ antagonism. Brain levels of nor-BNI and JDTic peaked within 30 min, but while nor-BNI was largely eliminated within hours, JDTic declined gradually over a week. Brain uptake of GNTI was too low to measure accurately, and higher doses proved lethal. None of the drugs were highly lipophilic, showing high water solubility (> 45 mM) and low distribution into octanol (log D7.4 < 2). Brain homogenate binding was within the range of many shorter-acting drugs (>7% unbound). JDTic showed P-gp-mediated efflux; nor- BNI and GNTI did not, but their low unbound brain uptake suggests efflux by another mechanism. Conclusions The negative plasma concentration-effect relationship we observed is difficult to reconcile with simple competitive antagonism, but is consistent with desensitization. The very slow elimination of JDTic from brain is surprising given that it undergoes active efflux, has modest affinity for homogenate, and has a shorter duration of action than nor-BNI under these conditions. We propose that this persistence may result from entrapment in cellular compartments such as lysosomes.