The environment is rarely constant and organisms are exposed to temporal and spatial variations that impact their life histories and inter-species interactions. It is important to understand how such variations affect epidemiological dynamics in host-parasite systems. We explored effects of temporal variation in temperature on experimental microcosm populations of the ciliate Paramecium caudatum and its bacterial parasite Holospora undulata. Infected and uninfected populations of two P. caudatum genotypes were created and four constant temperature treatments (268C, 288C, 308C and 328C) compared with four variable treatments with the same mean temperatures. Variable temperature treatments were achieved by alternating populations between permissive (238C) and restrictive (358C) conditions daily over 30 days. Variable conditions and high temperatures caused greater declines in Paramecium populations, greater fluctuations in population size and higher incidence of extinction. The additional effect of parasite infection was additive and enhanced the negative effects of the variable environment and higher temperatures by up to 50 per cent. The variable environment and high temperatures also caused a decrease in parasite prevalence (up to 40%) and an increase in extinction (absence of detection) (up to 30%). The host genotypes responded similarly to the different environmental stresses and their effect on parasite traits were generally in the same direction. This work provides, to our knowledge, the first experimental demonstration that epidemiological dynamics are influenced by environmental variation. We also emphasize the need to consider environmental variance, as well as means, when trying to understand, or predict population dynamics or range.