We test whether data from laboratory bioassays can be used to predict zooplankton responses during a whole—lake experiment using two rotifers, Keratella cochlearis and Keratella taurocephala. The acidification experiment was conducted in Little Rock Lake, Wisconsin, USA, which was divided into a reference basin maintained at a natural pH near 6.1 and a treatment basin which was acidified in 2—yr stages to pH values of 5.6, 5.2, and 4.7. Laboratory assays examined the effect of pH on reproduction under varied food conditions and survivorship without food. In the lake, the two rotifers showed strong and opposite responses to acidification: K. cochlearis decreased in abundance while K. taurocephala increased. In the laboratory bioassays, neither species was sensitive to pH when food conditions yielded high reproductive rates. When food was limited, K. cochlearis exhibited lower survivorship and a trend towards lower reproductive rates at lower pH. With limited food, K. taurocephala survivorship was either unaffected by pH or higher at high pH and its reproduction was slightly higher at intermediate pH. In situ experiments revealed that food conditions in the treatment basin lowered reproduction by K. cochlearis, indicating that a combined effect of food and pH caused its population decline. Neither food nor pH could explain the increase in K. taurocephala, which appeared to be linked to a reduction in its predators at lower pH. Overall, our analyses revealed substantial discrepancies between laboratory bioassays and in—lake responses. This was particularly the case for K. taurocephala, for which assays predicted no changes or a decline in abundance rather than the marked increase that actually occurred. Our results suggest that caution should be used in extending results from laboratory assays to natural ecosystems.