This paper investigates events in which cyclists perceive a cycling crash is narrowly avoided (henceforth, a near miss). A cohort of 2038 adult transport and recreational cyclists from New South Wales (Australia) provided self-reported prospectively collected data from cycling diaries to allow the calculation of an exposure-based rate of near misses and investigation of near miss circumstances. During 25,971days of cycling, 3437 near misses were reported. For a given time cycling, cyclists who rode mainly for transport (compared with those who rode mainly for recreation), and cyclists with less experience (compared to those with more experience) were more likely to report a near miss; older cyclists (60+ years) were less likely to report a near miss than younger cyclists (25-59 years). Where type of near miss was recorded, 72.0% involved motor vehicles, 10.9% involved pedestrians and 6.9% involved other cyclists. Results indicate some similarities between near misses and crashes reported by this cohort during the same reporting period. A bias toward reporting near misses with motor vehicles was suggested, which likely reflects cyclists' perceptions that crashes involving motor vehicles are particularly serious, and highlights their impact on perceived safety. Given the relative rarity of crashes, and the limited breadth and depth of administrative data, collection of near miss data may contribute to our understanding of cycling safety by increasing the volume and detail of information available for analysis. Addressing the causes of near misses may offer an opportunity to improve both perceived and actual safety for cyclists.