Evolutionary expectations about the virulence of parasites (i.e., the parasite-induced mortality rate of the host) often focus solely on the within-host transmission stage, overlooking the time spent between hosts and variations in transmission cycles. Moreover, the parasite growth rate within the host is closely linked to virulence. We here suggest that a simplified view of transmission and parasite evolution makes predicting how virulence will evolve difficult. We illustrate our ideas with a parasite with a simple life cycle, the microsporidian Vavraia culicis, which infects the mosquito Anopheles gambiae. We selected the parasite over six host generations for early or late host transmission, corresponding to shorter or longer time within the host. Selecting for late transmission increased their exploitation of the host, resulting in higher host mortality and a shorter life cycle with rapid infective spore production, comparatively to selection for early transmission. In response, hosts infected with late-selected spores shortened their life cycle and shifted to earlier reproduction. Using different host harm metrics, we demonstrate and discuss the pros and cons of using different it as measures of virulence. These and other findings emphasize the importance of considering the entire transmission cycle in studies of parasite evolution and raise concerns about how host density and social settings might influence virulence evolution.