A technique to identify the host crop of Heliothis moths by analysis of elemental content has been progressively developed over several seasons (CRC Project CS13L, Fin 1986). In this grant period we finalised statistical analyses of the existing database for individuals from known hosts and have developed a procedure for allocating unknown individuals collected in cotton crops to probable hosts (developed by Dr. Richard Motton CSRO Biometry Unit). The procedure (based on Newton's method with gradient matrix calculated by formula) allows for moths to be allocated to one of five major host crops or to a class derived from none of the major crops and includes a facility to use only a subset of possible hosts in an allocation run. The procedure gives acceptably accurate estimates of the proportion of a sample of moths derived from major hosts, though to date it has been validated using the same database as used to produce it. Further data for individuals from known hosts is being collected this season for validation purposes. We have commenced processing elemental data for unknowns collected in 1985/86 using the procedure but have encountered a problem which may be due to dust contamination of field caught moths. The elements Fe, P, and Mn are the most important in discriminating moths from different crops (accounting for 85-90% of the variation), but most of the moths taken from traps show levels of Fe (and Al and Si) outside the range of any of those reared from crops, while levels of other elements are similar to those for moths reared on crops. Elevated concentrations of Fe, Al and Si are characteristic of sample being contaminated with soil, which may easily occurring pheromone traps from which the moths are collected.