Simultaneous effects of host and parasite in determining quantitative traits of infection have long been neglected in theoretical and experimental investigations of host-parasite coevolution with the notable exception of gene-for-gene resistance studies. A cross-infection experiment, using five lines of the plant Arabidopsis thaliana and two strains of its oomycete pathogen Hyaloperonospora parasitica, revealed that three traits traditionally considered those of the parasite (number of infected leaves, transmission success, and time until 50% transmission), differed among specific combinations of host and parasite lines, being determined by the two protagonists of the infection. However, the two parasite strains did not differ significantly for most measured phenotypic traits of the infection. Globally, transmission increased with increasing virulence among the different host-parasite combinations, as assumed by most models of evolution of virulence. Surprisingly, however, there was no general relationship between parasite and host fitness, estimated respectively as transmission and seed production. Only one of the two strains showed the expected significant negative genetic correlation between these two variables. Our results thus highlight the importance of taking into account both host and parasite genetic variation because their interaction can lead to unexpected evolutionary outcomes.