We assess the environmental hazard of 50 pesticides used in Italy by m
eans of Hasse diagrams, a method based on graph theory. The criteria w
e use for ranking are persistence, and the physical-chemical propertie
s, vapour pressure and water solubility, and yearly usage. When only t
he physical-chemical properties plus persistence are used to assess en
vironmental hazard of pesticides in soils, eleven out of the 50 compou
nds studied here, methylbromide, bentazone, dalapon, diquat, linuron,
mancozeb, metham-Na, TCA, metolachlor, paraquat, and simazine are cons
idered potentially hazardous for the combination of long persistence i
n soil, high water solubility and low vapour pressure. Alachlor, atraz
ine, chloridazon, terbuthylazine and ziram are also a problem of conce
rn because of their high loadings. To test whether the theoretical ran
king is realistic, the list of identified compounds was compared with
the results of monitoring studies carried out in the River Po. The ass
umption for this comparison is that, if the ranking method is correct,
the probability of finding chemicals identified as hazardous should b
e higher than the probability of finding less hazardous chemicals. Che
micals ranked lower have less probability of being round both because
of lower usage and because they are less persistent and/or less leacha
ble. Results are quite encouraging since seven pesticides identified b
y our ranking method as most hazardous, alachlor, atrazine, bentazone,
linuron, metolachlor, simazine and terbuthylazine of the 8 analyzed f
or (previous plus TCA) were round, a success ratio of 88%. Results for
all the other chemicals are presented in the paper. The second purpos
e of this study was the identification of the most important criteria
to assess the chemicals; this assessment was performed using a matrix
W. We concluded that the elimination of the criterion ''usage'' affect
s ranking more than the elimination of water solubility. However, none
of the criteria, water solubility, vapour pressure, persistence and y
early usage can be eliminated, too much information would be lost if t
hey were omitted. This conclusion is consistent with our decision to u
se only few criteria to rank the chemicals, criteria that are deemed t
o be independent of each other. Copyright (C) 1996 Elsevier Science Lt
d.