Masking evaluators is a complex process, calling upon the creativity of sci
entific investigators and the unrelenting attention of project directors an
d data collectors. This article reviews research supporting the importance
of masking to reduce the effects on outcome of evaluator bias, especially g
iven the evidence of strongly held beliefs about the effects of prenatal co
caine exposure. We discuss the particular difficulty of totally masking exa
miners who are blind to drug status yet aware of the nature of the study, g
iven an association of drug use and sociodemographic and lifestyle conditio
ns that evaluators may know or observe. Finally we suggest techniques that
may be used to facilitate masking. As part of a longitudinal study of the e
ffects of prenatal cocaine exposure, we asked blinded evaluators, after com
pleting the Brazelton Neonatal Behavioral Assessment Scale, to guess whethe
r or not they thought each infant had been cocaine-exposed. The 276 infants
were evaluated at birth in a research room without their parents so that t
he only information available to examiners was what could be observed about
the infant Those infants whom examiners guessed to be cocaine-exposed were
significantly smaller by birthweight, were more likely to be black, and ha
d poorer Brazelton performance. The guesses were correct slightly more ofte
n than chance (56 percent; p = .050) because of an actual, though weak, rel
ationship between size and performance. In short the guesses of examiners d
emonstrated a bias about the relationship of poorer performance and cocaine
exposure, supporting the critical importance of our efforts to maintain ma
sking in evaluations.