1. Pieris brassicae lays large batches of eggs (up to 150) and the cat
erpillars that hatch frequently have to migrate in search of new plant
s before completing development. Some of the costs and benefits of thi
s behaviour for the caterpillars are investigated. 2. A model is devel
oped to show how the degree of inhibition of plant growth due to larva
l feeding can affect the extent to which more food is available to a g
roup of larvae feeding on two plants sequentially than to two groups,
each of half the size, feeding on two plants simultaneously. 3. In a f
ield experiment, brussels sprouts plants were infested with 0, 5, 40 o
r 100 first instar P. brassicae caterpillars. The presence of caterpil
lars was found to have an inhibitory effect on plant growth. No effect
s of caterpillar group size on mortality during the first three instar
s or on parasitism by Cotesia glomerata were found. 4. In a second fie
ld experiment, fourth and fifth instar caterpillars were released eith
er 0.5 m or 2.5 m away from host plants. After 72 h, 81% of those rele
ased 0.5 m away had located a plant, but only 16% of those released 2.
5 m away had done so. The results were not significantly affected by t
he age of the caterpillars used or by the density of background vegeta
tion. 5. It is suggested that laying more eggs than can be supported b
y a single host plant may be beneficial to female butterflies if the l
ocal density of host plants is high. After exhausting their natal host
plant, caterpillars can then exploit the uninhibited growth of neighb
ouring and previously uninfested plants with only a small cost to migr
ation. Under such circumstances, butterflies might be expected to trea
t clumps of host plants, rather than individuals, as the unit of resou
rce over which oviposition decisions should be made.