As a result of routine monitoring data on carrots generated in the 1990s indicating MRL exceedances, further studies showed that residues of organophosphorus compounds in individual roots could vary up to 25 times the levels found in composite samples. Additional work found that this phenomenon also extended to other crops including apples, peaches and celery. Variability (defined as the highest residue level found in any one crop item divided by the level found in a composite sample from the same batch) of up to 34 times mean values was found in one batch of plums. Conventional deterministic methods used in consumer assessments were likely to give gross overestimates of short-term exposure because of the assumptions employed. This led to the development of probabilistic models which made the best use of all available information and was capable of indicating percentages of consumers that could exceed relevant toxicological end points. This indicated that there was unlikely to be serious health effects as a result of ingesting these residues and has subsequently proved to be a useful aid to regulatory decision making.