In short-term earthquake risk forecasting, the avoidance of false alarms is of utmost importance to preclude the possibility of unnecessary panic among populations in seismic hazard areas. Unusual animal behaviour prior to earthquakes has been reported for millennia but has rarely been scientifically documented. Recently large migrations or unusual behaviour of amphibians have been linked to large earthquakes, and media reports of large frog and toad migrations in areas of high seismic risk such as Greece and China have led to fears of a subsequent large earthquake. However, at certain times of year large migrations are part of the normal behavioural repertoire of amphibians. News reports of “frog swarms” from 1850 to the present day were examined for evidence that this behaviour is a precursor to large earthquakes. It was found that only two of 28 reported frog swarms preceded large earthquakes (Sichuan province, China in 2008 and 2010). All of the reported mass migrations of amphibians occurred in late spring, summer and autumn and appeared to relate to small juvenile anurans (frogs and toads). It was concluded that most reported “frog swarms” are actually normal behaviour, probably caused by juvenile animals migrating away from their breeding pond, after a fruitful reproductive season. As amphibian populations undergo large fluctuations in numbers from year to year, this phenomenon will not occur on a yearly basis but will depend on successful reproduction, which is related to numerous climatic and geophysical factors. Hence, most large swarms of amphibians, particularly those involving very small frogs and occurring in late spring or summer, are not unusual and should not be considered earthquake precursors. In addition, it is likely that reports of several mass migration of small toads prior to the Great Sichuan Earthquake in 2008 were not linked to the subsequent M = 7.9 event (some occurred at a great distance from the epicentre), and were probably co-incidence. Statistical analysis of the data indicated frog swarms are unlikely to be connected with earthquakes. Reports of unusual behaviour giving rise to earthquake fears should be interpreted with caution, and consultation with experts in the field of earthquake biology is advised.