Purpose. This study retrospectively compared subjects from three unrel
ated studies using eight algorithms to stage exercise behavior. Subjec
ts and settings. Study One included 936 employees involved in a smokin
g cessation study at four worksites-a medical center; retail store, ma
nufacturing firm, and a government agency. Study Two included 19,212 m
embers of a New England HMO; and Study Three included a convenience sa
mple of 327 adult New Englanders. Measures. The eight algorithms used
different descriptions of stages based on the transtheoretical model,
as well as different definitions of exercise and response formats. Res
ults. Algorithms using longer; more precise definitions of exercise re
sulted in larger numbers of subjects being staged in precontemplation
and contemplation in comparison to algorithms using shorter definition
s, which tended to stage subjects in preparation and action. Maintenan
ce was the most and preparation the least consistently described stage
across algorithms. Conclusions. Alteration of the descriptions of sta
ge and the definition of exercise has consequences for the staging of
subjects. Definitions need to be explicit, stating all parameters need
ed to meet criterion, and subjects must be able to assess themselves.
Either a 5-Choice or a true/false response format is effective in asse
ssing stage.