Abstract Natural product insecticides used in this experiment were Bacillus thuringiensis var. san diego (MYX 1806), B. thuringiensis var. tenebrionis (Trident and BK-102) and azadirachtin (Margosan-O) from the neem tree, Azadirachta indica. Plots were planted 15 May at the Research Farm, Red River Valley Potato Growers Association, Grand Forks, N.D. Plots were 6 rows wide, 16.8 (55 ft) long, with 91 cm (36 inches) spacing between rows and 30 cm (12 inches) between plants. The two center rows of each plot were planted with certified 'Red Pontiac' seed pieces, the remaining four rows were planted with certified, B-size, 'Russet Burbank.' Plots were separated on all sides by 1.8 m (6 ft) of cultivated soil. A completely randomized plot design with 4 replications was used. Plots were scouted daily from CPB egg masses beginning 21 Jun. When egg masses were found these were flagged and checked daily for hatch. Timing of first spray application was based on egg hatch. Three basic spray schedules were used: first application at 30% hatch of flagged egg masses and subsequent application on either a 7 d or 10 d schedule, or first application at 30% hatch of all eggs (unflagged) laid on or before that sampling date and subsequent applications on a 7 d schedule. Plots were sprayed 1-3 times for each generation of CPB larvae: 29 Jun (30% hatch of flagged eggs—1st generation) 2 Jul (30% total hatch—1st generation), 5,9, 12, 16, 19 and 31 Jul (30% hatch of flagged eggs—2nd generation), 6 Aug (30% total hatch—2nd generation), 10, 13 and 14 Aug. Insecticidal sprays were applied using a tractor-mounted boom sprayer, with a single Teejet nozzle over each row. Sprays were applied in 150 liters water/ha (40 gal/ acre) at 275 kPa (40 psi). All data were collected from the two center rows of each plot. CPB larvae (small = 1st and 2nd instar, large = 3rd and 4th instar) were counted on 20 randomly selected whole plants per plot, PLH nymphs were counted on 35 mid-plant leaves per plot. Leaf loss to defoliation (by CPB) and hopperbum (injury caused by PLH) were estimated visually (by 2 individuals per sampling date) as percent foliage lost. CPB were counted on 4, 10 and 17 Jul, and 1, 8 and 17 Aug; defoliation was estimated on 10, 17 and 24 Jul, 1, 8 and 15 Aug; PLH were counted on 4, 10, 17 and 24 Jul, 1, 8 and 17 Aug; hopperbum was estimated on 1, 8 and 15 Aug. Insect data and plant injury ratings were expressed as mean insect numbers or plant injury × number of days elapsed (= Area under Curve). AUC for CPB was calculated separately for 1st and 2nd generations, and then summed. Yields were obtained 2 Oct by harvesting one center row from each plot. Two way analysis of variance was used to test for treatment differences in pest densities, plant injury and yield. The 1990 growing season at Grand Forks was characterized by severe drought from late Jun through harvest.