Cholinesterase (ChE) levels (Ellman method) were monitored in 90 subjects (69 males and 21 females) exposed to carbamate and organophosphate pesticides (78 agricultural workers and 12 pesticide vendors). Pre-exposure baseline values of plasma and red blood cell cholinesterase activities were defined for each subject with two blood samples (23 workers) or three blood samples (59 workers) taken almost thirty days after the last exposure. After control of intra-individual variation, 8 subjects with only one pre-exposure value and 13 with a coefficient of variation above 30% were excluded. For the other 59 subjects, the intra-individual variation of erythrocyte ChE (16%) was similar to the inter-individual one (15%), whereas the inter-individual variation of plasma ChE (21%) was higher than the intra-individual one (14%). Laboratory variation for plasma ChE measurements was 8%. Baseline values were analyzed (ANOVA) for sex, age, task and hour and season of sampling. Both erythrocyte and plasma enzymes, corrected for hematocrit, were lower in females. Plasma cholinesterase activity was lower in "re-entry" agricultural workers and in pesticide vendors. Post-exposure cholinesterase activity was measured in 54 workers within a few (1-21) days after last handling. Average relative reduction was 15.2% (95% C.I. = 4.9%-25.5%) in erythrocyte cholinesterase activity and 29.1% (95% C.I. = 18.2%-40.1%) in plasma cholinesterase activity. The one-way variance analysis showed marked plasma ChE reduction in mixers, loaders and appliers (36%, 95% C.I. = 24%-48%) and in parathion handlers (35%, 95% C.I. = 21%-49%. No significant reduction in blood cell cholinesterase activity in relation to task and to pesticide handled was observed. We conclude that the intra-individual variations of the baseline values were higher for three repetitions (88% and 84% of the population were within a variability of less than 30%, for AChE and for ChE respectively) than for two repetitions (91% and 88% of the population were within 30% of variability for AChE and for ChE respectively). The figures show a greater sensitivity of plasma ChE activity in acute exposure, probably due to a poor reliability in detection of erythrocyte ChE by local laboratories. The maximum reduction (38%, 95% C.I. = 22%-53%) in plasma ChE activity was observed within six days of the last exposure in loaders and appliers.