Positive allosteric modulators (PAMs) of metabotropic glutamate receptor subtype 5 (mGlu(5)) have emerged as an exciting new approach for the treatment of schizophrenia and other central nervous system (CNS) disorders. Of interest, some mGlu(5) PAMs act as pure PAMs, only potentiating mGlu(5) responses to glutamate whereas others [allosteric agonists coupled with PAM activity (ago-PAMs)] potentiate responses to glutamate and have intrinsic allosteric agonist activity in mGlu(5)-expressing cell lines. All mGlu(5) PAMs previously shown to have efficacy in animal models act as ago-PAMs in cell lines, raising the possibility that allosteric agonist activity is critical for in vivo efficacy. We have now optimized novel mGlu(5) pure PAMs that are devoid of detectable agonist activity and structurally related mGlu(5) ago-PAMs that activate mGlu(5) alone in cell lines. Studies of mGlu(5) PAMs in cell lines revealed that ago-PAM activity is dependent on levels of mGlu(5) receptor expression in human embryonic kidney 293 cells, whereas PAM potency is relatively unaffected by levels of receptor expression. Furthermore, ago-PAMs have no agonist activity in the native systems tested, including cortical astrocytes and subthalamic nucleus neurons and in measures of long-term depression at the hippocampal Schaffer collateral-CA1 synapse. Finally, studies with pure PAMs and ago-PAMs chemically optimized to provide comparable CNS exposure revealed that both classes of mGlu(5) PAMs have similar efficacy in a rodent model predictive of antipsychotic activity. These data suggest that the level of receptor expression influences the ability of mGlu(5) PAMs to act as allosteric agonists in vitro and that ago-PAM activity observed in cell-based assays may not be important for in vivo efficacy.