The development of a new data set on the 1800-1990 average nominal tar
iff rates of five of the system's most important traders (Britain, Fra
nce, the United States, Germany and Japan) permits a new empirical exa
mination of alternative explanations of protectionism. Seven models, w
ith various emphases on hegemony, business cycles, war shocks and econ
omic growth, are used to generate nine schedules of expected changes i
n the direction of tariff rates. Correlations of the expected schedule
s with the observed directional shifts demonstrate clearly the timing
superiority of explanations based on long wave and external shock emph
ases. Yet even the strongest performing models do less well at levels
of analysis other than the system level. Greater model elaboration, it
is concluded, is needed to better account for processes at work in di
fferent subsystems.