Biological messengers can be "caged" by adding a single photosensitive group that can be photolyzed by a light flash to achieve spatially and temporally precise biochemical control. Here we report that photolysis of a double-caged form of the second messenger inositol 1,4,5-trisphosphate (IP3) triggers focal calcium release in Purkinje cell somata, dendrites, and spines as measured by two-photon microscopy. In calbindin knock-out Purkinje cells, peak calcium increased with flash energy with higher cooperativity for double-caged IP3 than for conventional single-caged IP3, consistent with a chemical two-photon effect. Spine photolysis of double-caged IP3 led to local calcium release. Uncaging of glycerophosphoryl-myo-inositol 4,5-bisphosphate (gPIP2), a poorly metabolizable IP3 analog, led to less well localized release. Thus, IP3 breakdown is necessary for spine-specificity. IP3- and gPIP2-evoked signals declined from peak with similar, slow time courses, indicating that release lasts hundreds of milliseconds and is terminated not by IP3 degradation but by intrinsic receptor dynamics. Based on measurements of spine-dendrite coupling, IP3-evoked calcium signals are expected to be at least 2.4-fold larger in their spine of origin than in nearby spines, allowing IP3 to act as a synapse-specific second messenger. Unexpectedly, single-caged IP3 led to less release in somata and was ineffective in dendrites and spines. Calcium release using caged gPIP2 was inhibited by the addition of single-caged IP3, suggesting that single-caged IP3 is an antagonist of calcium release. Caging at multiple sites may be an effective general approach to reducing residual receptor interaction.