A packed bed bioreactor, with 170 ml glass bead carriers and 130 ml medium, was tested for the removal of the endocrine disrupter, nonylphenol, with a Sphingomonas sp. The bioreactor was first continuously fed with medium saturated with nonylphenol in an attempt to simulate groundwater pollution. At best, nonylphenol was degraded by 99.5% at a feeding rate of 69 ml h(-1) and a removal rate of 4.3 mg nonylphenol day(-1), resulting in a 7.5-fold decrease in effluent toxicity according to the Microtox. The bioreactor was then fed with soil leachates at 69 ml h(-1) from artificially contaminated soil (1 g nonylphenol kg(-1) soil) and a real contaminated soil ( 0.19 g nonylphenol kg(-1) soil). Nonylphenol was always completely removed from the leachates of the two soils. It was removed by 99% from the artificial soil but only 62% from real contaminated soil after 18 and 20 d of treatment, respectively, showing limitation due to nonylphenol adsorption.