Knowledge about the ability of freshwater unionid mussels to recover from physical disturbance is important to their conservation and management. Threatened species may be disturbed by relocation to refugia as a conservation measure,and some species are disturbed by size- and species-selective harvesting of shells for use in the production of cultured pearls. The activity of freshwater unionid mussels generally decreases with water temperature, but intra- and interspecific differences in the frequency and distribution of recovery behaviors following disturbances at specific water temperatures have not been previously quantified. We observed righting, moving, and burrowing behavior of 4 mussel species, Amblema plicata plicata, Potamilus alatus, Fusconaia flava, and Lampsilis cardium, at 3 water temperatures (7, 14 and 21 degrees C). The temporal frequency (intensity) and times-to-1st-event of behaviors were analyzed using proportional hazards models. Righting events and consecutive movements occurred at different intensities among temperatures and species. For righting, intensity increased by 8%/degrees C within the range of 7-21 degrees C. Subsequent movements increased in intensity by 10%/degrees C. Amblema plicata was the slowest to respond, and had an intensity of turning upright only 27% of that for P. Alatus. The intensities of movements for A. Plicata and E. Flava were 16% of those for P. Alatus. Lampsilis cardium righted themselves most quickly, and had an intensity of righting 124% of that for I! alatus. The distribution of the 3 behaviors among treatment groups at 1 wk was analyzed with a proportional odds model. The distribution of righting, moving, and burrowing 1 wk after disturbance was described entirely by high-order interactions in our proportional odds model. Therefore, that model revealed little interpretable pattern in the endpoint data and it was less sensitive than our analysis of time-to-event data for measuring the effects of disturbance. We attributed the difference in sensitivity between the 2 models to the greater information content of time-until-event data. For similar studies of occurrences of key events, times to events should be recorded and interpreted whenever feasible and consistent with study objectives, Our results suggest that water temperature has an important effect on the outcome of mussel conservation projects and commercial harvesting activities. Our modeling approach, applied to other species, could help guide decisions about which species can safely be disturbed and the optimal seasonal timing of those disturbances.