The measurement of peak expiratory flow (PEF) has been applied in a nu
mber of different settings since Wright developed a convenient measuri
ng device.22 Although it has not proved as reliable as forced expirato
ry volume in 1 second (FEV1) measured by spirometry, the ease of PEF m
easurement and the current availability of inexpensive, portable instr
uments have led to increasing use of this measure in clinical evaluati
ons of ambulatory patients and in population studies to assess environ
mental and occupational hazards. In addition, patterns determined from
serial measurements of PEF are now accepted as the least invasive met
hod to identify individuals with airway reactivity associated with occ
upational exposures in both clinical and epidemiologic settings.3 As t
he application of PEF becomes more widespread, the need arises for sta
ndardization of the testing protocol and interpretation of test result
s. The American Thoracic Society (ATS) has published extensive informa
tion regarding the standardization of the FEV1 and forced vital capaci
ty (FVC) measured on a spirometer.1,2 These ATS recommendations are ba
sed on an extensive literature on spirometry testing, including studie
s of learning effects, instrument error, and sources of intrasubject v
ariation due to diurnal, day-to-day, and within-test session differenc
es. For these pulmonary function tests, the ATS statements provide det
ailed recommendations concerning equipment specifications and validati
on, testing protocol, and interpretation of test results. These statem
ents also include recommendations for the number of efforts that ought
to be performed at a test session, specification of an ''acceptable''
effort, and criteria for judging the reproducibility of the measureme
nt. By contrast, there is relatively little in the literature addressi
ng how best to measure PEF. No studies comparable to the Epidemiology
Standardization Project8 have examined the optimal frequency, instrume
nt precision, and appropriate standards of reproducibility of the test
. It is current practice for measuring PEF in both field and clinical
settings to record three blows and report the maximum. There is sporad
ic application of a reproducibility criterion of 20 L/min (see Burge,
pp 279-294 in this issue).18 Because PEF, particularly serial measures
of PEF, are likely to see a far broader application in the assessment
of acute respiratory hazards, a closer examination of the measurement
characteristics of PEF is warranted. A review of the literature was u
nder-taken to identify what is known concerning the variance of repeat
ed trials of PEF within a single test session (precision) for the PEF
meters in common use. Although we found frequent reference to the peak
flow as a more variable measurement than other pulmonary function par
ameters, e.g., FEV1, this conclusion was generally based on comparison
s of diurnal variation. There was also a literature addressing the acc
uracy of various PEF meters relative to measurements on the pneumotach
ograph. There was virtually no literature, however, describing the var
iability within a test session (precision). Although knowing that inst
ruments measure maximum flow achieved during a forced expiratory effor
t without systematic error is important, documenting precision is also
important. This is particularly true since short-term changes in PEF
over time have become the most interesting applications of PEF, making
relatively small differences in flow rates important to measure preci
sely (see Eisen et al., pp 265-277, in this issue). The absence of a l
iterature relevant to precision led us to examine data from one of our
own large worker population studies with repeated trials recorded at
each test session. Here, we present a summary of the limited literatur
e on precision along with results from that study.