Arsenic concentrations were measured in 992 drinking water samples collecte
d from New Hampshire households using online hydride generation ICP-MS. The
se randomly selected household water samples contain much less arsenic than
those voluntarily submitted for analysis to the New Hampshire Department o
f Environmental Services (NHDES). Extrapolation of the voluntarily submitte
d sample set to all New Hampshire residents significantly overestimates ars
enic exposure. In randomly selected households, concentrations ranged from
<0.0003 to 180 mu g/L, with water from domestic wells containing significan
tly more arsenic than water from municipal sources. Water samples from dril
led bedrock wells had the highest arsenic concentrations, while samples fro
m surficial wells had the lowest arsenic concentrations. We suggest that mu
ch of the groundwater arsenic in New Hampshire is derived from weathering o
f bedrock materials and not from anthropogenic contamination. The spatial d
istribution of elevated arsenic concentrations (>50 mu g/L) correlates with
Late-Devonian Concord-type granitic bedrock. Field observations in the reg
ion exhibiting the highest groundwater arsenic concentrations revealed abun
dant pegmatite dikes associated with nearby granites. Analysis of rock dige
sts indicates arsenic concentrations up to 60 mg/kg in pegmatites, with muc
h lower values in surrounding schists and granites. Weak acid leaches show
that approximately half of the total arsenic in the pegmatites is labile an
d therefore can be mobilized during rock-water interaction.