In the present paper a new shrinkage estimator of relative risk, usefu
l in disease mapping, is compared with the empirical and full Bayes es
timators, using death certificate data for lung cancer (males and fema
les) and breast cancer (females) 1982-1988 in the Emilia-Romagna regio
n (Italy). The estimates are obtained averaging the relative risks obt
ained from recursive partitioning of the set of observed values. The n
umber of partitions is fixed in advance and behaves as a smoothing par
ameter, the lower the number the higher the degree of shrinkage toward
the overall mean of the observed values.