The North American migratory grasshopper Melanoplus sanguinipes (Fabricius) has engaged in spectacular outbreak swarms several times in the 20th century, and substantial numbers of individuals in many populations perform long-distance flights even in non-outbreak years. In the field, migratory flights typically begin under specific conditions of light, temperature, and windspeed that can be simulated in laboratory tethered-flight experiments. There is individual variation in migratory tendency, both within and between populations, that appears to have a substantial genetic basis; movements of grasshopper populations are therefore influenced by both environmental and internal cues. Previous work has found that (1) tethered-flight performance in the laboratory is correlated with migratory tendency in natural populations; and (2) tethered-flight performance is bimodally distributed, because individuals tend either to not fly (or fly for only a few minutes), or to fly for much longer than one hour. In the present work, we analyze migratory tendency as a threshold trait, using performance of 60 - 80 min of tethered-flight in at least one of three trials as diagnostic of a prospective migrant. In a laboratory population derived from a migratory Arizona population, heritability of migratory tendency is estimated to be in the range 0.3 - 0.5. In males, artificial selection on the frequency of flights 60 - 80 min in duration increases both the frequency and the mean duration of flights > 60 min. Previous work suggests that while carbohydrate is the principal fuel for short flights and flight initiation, lipid is the principal fuel for long-duration flight. Therefore, in this study we examine the correlated response of characters relating to lipid metabolism to artificial selection on migratory tendency. These characters include lipid reserves as percent of body mass, thoracic volume, relative size of thoracic lipid reserves, and amount of adipokinetic hormone (AKH) stored in the corpora cardiaca.