Invasive species often have instable population dynamics and are known to collapse or oscillate heavily after passing through the initial lag/growth phases. Long-term data-series documenting these fluctuations are however rare. We use long-term (starting in the early 1960s), semi-quantitative data on the invasive signal crayfish (Pacifastacus leniusculus), capturing its population development after introduction in 44 Swedish lakes. In total 18 (41 %) of these populations had experienced a collapse. A stepwise discriminant function analysis including 20 different ecological or physicochemical characteristics identified three variables explaining collapses in the following order: stocking year, population age and mean air temperature. Populations stocked in the 1980s were more likely to collapse than populations stocked in the 1970s. Lakes with collapses were located in areas with 0.4 A degrees C higher yearly mean air temperatures than the still viable populations. Collapses also depended on the time phase of the population and started to occur 12 years after stocking and were most frequent in the interval 16-20 years after stocking and after 11-15 years duration of the established phase with harvestable densities. An analysis of prevalence and pathogen load of Aphanomyces astaci was conducted in eight of the studied populations. A. astaci was present in all populations but neither the level of prevalence nor the pathogen load in infested specimens differed significantly between lakes with collapses and lakes without. Our results highlight the potential sensitivity and instability of introduced crayfish. The importance of density-dependence and temperature suggest that both climate variability and/or fisheries can influence these processes.