Pesticide residue levels (36 pesticides and some of their metabolites) were determined in the individual units taken from large samples of apples, bananas, celery, kiwi fruit, oranges, peaches and nectarines, pears, plums, potatoes, and tomatoes. The 65 large samples (generally about 100-110 units, but only 45 units of celery) were purchased at retail or wholesale outlets in the UK. The lots from which the samples were drawn originated from 17 different countries. Average concentrations in the samples were in the approximate range 0.002-2 mg kg(-1). Unit-to-unit variability factors (97.5th percentile mg kg(-1)/average mg kg(-1)), for the pesticide/product combination data sets in which >10% of samples contained measurable residues (n = 106), were in the range 1.4-9.6 (11.1 based on a value of zero for data below reporting limits). Analytical variance contributed only a small proportion (up to 11%) to the overall variance of the 106 data sets. There was no evidence of a relationship between the variability factor and the commodity, country of origin, residue concentration or the physicochemical characteristics of the pesticide. The extent of variability appears to be determined at or about the time of pesticide application. Taking non-detectable residues as half the reporting limits, the frequency distribution of variability factors was approximately log-normal, with a geometric mean of 3.4. The corresponding 95% probability limits of the variability factors were calculated to be 1.6 and 7.6.
