Serial peak expiratory flow (PEF) measurement has been increasingly used in the study of variable airway disease including asthma. In asthmatics, the amplitude of the circadian rhythm in PEF is exaggerated compared to normal. Most commonly, the amplitude has been calculated as the difference between the maximum and minimum flow rates relative to either the maximum or mean flow rates in a day. Different studies have adopted different number of measurements at different times of day. We examined the effect of the time and number of PEF measurements in detecting the evidence for variable airway disease in a group of workers. Mini-Wright peak flow meters were given to 29 subjects working day shift from Monday to Friday in an insect-rearing facility. They were asked to record three blows every two hours while awake for 8 days. Each participant was checked daily by the study team to insure correct technique and maximal effort. On weekends, they were called at home to encourage compliance. Two workers had only 7 days of recording. All others completed 8 days of PEF measurement. Among 29 workers, 8 showed one or more days of >/= 20% variation in PEF (maximum minus minimum divided by maximum). Of the total 64 days of PEF measurement in these 8 workers, 16 days had >/= 20% diurnal variation. The effect of timing of PEF measurement was examined by calculating the number of days with >/= 20% variation after deleting the data at one specific time of day, keeping the rest of the data in the analysis. Deleting the 6 am measurements reduced the diurnal variation of 7 days. 8 pm, and 12 noon influenced 4 and 3 days each. Other times of day which changed the status of diurnal variation from >/= 20% to < 20% included 8 am, 2 pm, 6 pm, and 10 pm. The influence of number of PEF measurements was examined by keeping the data at specified times of day in the analysis. Analysis of two (6 am and 8 pm), three (6 am, 12 noon, and 8 pm), four (6 am, 12 noon, 2 pm, and 8 pm), and five (6 am, 12 noon, 2 pm, 8 pm, 10 pm) PEF measurements showed that at least five were needed to detect all 8 workers with excessive diurnal variation. Analysis using three and four measurements detected 5 and 7 workers each, while analysis with two was quite inadequate, detecting only 1 subject. These results emphasize the importance of early morning blows and the use of at least 3-4 measurements in detecting excessive diurnal variation of peak flow rates in working people.