We addressed several key questions about duration of bioassays, diagnostic concentration, and probit regression slope for response of diamondback moth, Plutella xylostella (L.), to the microbial insecticide, Bacillusthuringiensis Berliner. Data were obtained from 54 bioassays of larvae from seven field populations and eight laboratory colonies in Hawaii. Linear regression showed that LC50s at 96 and 120 h after exposure to B. thuringiensis were associated with LC50s at 24 and 48 h. Control mortality increased as time increased, but time had little effect on slope, standard error of slope, or 95% fiducial limits for LC50 A significant portion of the variation in LC50 was accounted for by variation in mortality at a single concentration. Linear and polynomial regression showed that slope did not vary as a simple or consistent function of LC50. Variation in LC50 across all strains of diamondback moth was >5,000-fold, yet variation in LC50 for repeated bioassays of single strains was