Hippocampal depolarization-induced suppression of inhibition (DSI) is a robust form of short-term synaptic plasticity. DSI is mediated by endocannabinoid signaling. Since this discovery, pinning down the endogenous cannabinoid receptor ligand that mediates DSI has been problematic. Blocking degradation of the endocannabinoid 2-arachidonoyl glycerol (2-AG) lengthens DSI, which seems to indicate that 2-AG mediates DSI. In contrast, pharmacological inhibition of the 2-AG-synthesizing enzyme diacylglycerol lipase (DAGL) has yielded conflicting results: DAGL inhibitors often fail to block hippocampal DSI. Recently, 2 studies seem to have cornered this problem using DAGL knockout mice. Hippocampal DSI is absent in DAGL-α knockout mice, pointing to a key role for 2-AG in DSI. However, these studies do not reconcile the discrepancy with pharmacological experiments. Here, we argue that the seeming contradiction between results from pharmacological and genetic approaches may be explained in several ways. We suggest that the contradiction may be resolved by taking a different perspective on endocannabinoid signaling: in some forms of endocannabinoid-mediated signaling endocannabinoids might not be necessarily produced "on demand" but presynthesized and stored until needed.