A comprehensive and effective framework is designed to index and mine large-scale neuron morphological databases. A novel retrieval method is proposed based on maximum inner product search and feature hierarchy. Its potential use cases are demonstrated for neuron identification and property analysis.