A high-resolution geospatial dataset of cyclist-involved crashes and segmented cycling infrastructure in the Guadalajara metropolitan area (2015-2024).

We release a geospatial, segment-level dataset linking 774 cyclist-involved crashes (2015-2024) in the Guadalajara Metropolitan Area (Mexico) to a uniformly segmented cycling network. The network is split into ∼20-m micro-segments, enabling the assignment of each crash to a unique segment identifier (IDSegmento) and the computation of a segment level event concentration variable (Concentracion_accidentes). All records are validated for spatial plausibility and key integrity; machine-readable metadata, schemas, codebooks, and checksums are included for full reproducibility. This structure facilitates fine-grained spatial analysis, hotspot detection, and model training without further geocoding. The dataset is suitable for research on cyclist safety, infrastructure planning, and policy evaluation, and can be replicated in other Latin American cities under similar data conditions.