Variation between samples of the Colorado potato beetle, Leptinotarsa decemlineata (Say), and between mortalities from some factors, on potato, were determined in long-term study plots at Merivale, Ontario. For each of six age intervals (spring adults; egg masses; larvae, Periods 1 and 2; pupal cells; and summer adults), inter-hill variance was the major source of sampling variance. Intra-hill and block differences were rarely significant. The most appropriate sample unit for the above-ground stages was the potato stalk and for pupal cells, the soil quadrant. Optimum allocation procedures indicated that not more than one sample unit per hill need be taken for the immature stages. For adults, the whole hill should be sampled. To estimate population means with acceptable limits of precision (10% standard error), it is recommended that 50 to 200 hills be sampled for the various age intervals.