The mitigating effect of increasing hardness on metal toxicity is reflected in water quality criteria in the United States over the range of 25-400 mgl(-1) (as CaCO(3)). However, waters in the arid west of the US frequently exceed 400 mgl(-1) hardness, and the applicability of hardness-toxicity relationships in these waters is unknown. Acute toxicity tests with Ceriodaphnia dubia were conducted at hardness levels ranging from approximately 300 to 1,200 mgl(-1) using reconstituted waters that mimic two natural waters with elevated hardness: (1) alkaline desert southwest streams (Las Vegas Wash, NV), and (2) low alkalinity waters from a CaSO(4)-treated mining effluent in Colorado. The moderately-alkaline EPA synthetic hard water was also included for comparison. Copper toxicity did not consistently vary as a function of hardness, but likely as a function of other water quality characteristics (e.g., alkalinity or other correlated factors). The hardness equations used in regulatory criteria, therefore, may not provide an accurate level of protection against copper toxicity in all types of very hard waters. However, the mechanistic Biotic ligand model generally predicted copper toxicity within +/-2X of observed EC(50) values, and thus may be more useful than hardness for modifying water quality criteria.