Proximity labeling (PL) enables covalent capture of biomolecules near a defined target in living systems, providing a route to map molecular proximity in situ. Conventional PL usually relies on a single targeting cue and may therefore lack the selectivity required to resolve defined molecular events or dynamic processes in complex biological systems. Conditional PL addresses this limitation by coupling labeling activity to additional inputs, including interaction-dependent enzyme reconstitution, localization of catalytic modules at cell–cell interfaces, light-regulated catalysis, and multiple orthogonal labeling reactions. These designs allow labeling to respond to specified spatial, temporal, or molecular conditions. We group conditional PL into four designs—split complementation, interface localization, light control, and multiple labeling—based on what activates labeling and what the labels reveal. We examine their use in studying protein complexes, trafficking, and intercellular communication. We also consider what each strategy encodes, the experimental boundaries of the resulting proximity information, and trade-offs among selectivity, labeling efficiency, temporal control, and compatibility with native biological contexts. Remaining challenges include expanding catalyst and probe diversity, improving in vivo compatibility, and integrating conditional PL with single-cell analysis, spatial omics, and high-resolution imaging.