During the building of a process-based simulation model for the epidemiology of the multicapsid nucleopolyhedrovirus of S. exigua (SeMNPV) in populations of Spodoptera exigua (Hübner) in greenhouse chrysanthemum, it was found that the effect of host plants had been under-rated. 'Missing links' included (i) the 'natural' background mortality of larvae of S. exigua in practical cropping conditions; (ii) the developmental rate of larvae of S. exigua on plant substrate in a glasshouse as compared to artificial medium in the laboratory; (iii) the validity of the results of dose-mortality and time-mortality bioassays conducted on artificial medium as compared to natural plant substrate; (iv) the distribution of inoculum released from deceased caterpillars over chrysanthemum leaves; and (v) the leaf visit rate of healthy caterpillars (as it affects horizontal transmission). Experiments were carried out to quantify these processes. Developmental rates of S. exigua larvae on greenhouse chrysanthemum were 36␕ower than on an artificial diet. The fraction survival during the first, second, third and fourth instar S. exigua larvae in greenhouse chrysanthemum was 0.60, 0.80, 0.88 and 0.95, respectively. Forty percent of the first instar larvae reached the fifth larval stage. Second instar S. exigua larvae reared on chrysanthemum were significantly more susceptible to SeMNPV than larvae reared on an artificial diet. The food source had no effect on the time to kill S. exigua larvae. Cadavers of second, third and fourth instar S. exigua larvae contaminated on average 1.4, 2.5 and 3.3 chrysanthemum leaves. Second to fourth instar S. exigua larvae visited 23 leaves per day and spent 1555␘f the time on the underside of leaves. The above information is of critical importance for a trustworthy simulation of the epidemiology of SeMNPV in chrysanthemum.