Damage detection techniques have been proposed to exploit changes in modal parameters and to identify the extent and location of damage in large structures. Most of such techniques, however, generally neglect the environmental effects on modal parameters. Such environmental effects include changes in loads, boundary conditions, temperature, and humidity. In fact, the changes due to environmental effects can often mask more subtle structural changes caused by damage. This paper examines a linear adaptive model to discriminate the changes of modal parameters due to temperature changes from those caused by structural damage or other environmental effects. Data from the Alamosa Canyon Bridge in the state of New Mexico were used to demonstrate the effectiveness of the adaptive filter for this problem. Results indicate that a linear four-input (two time and two spatial dimensions) filter to temperature can reproduce the natural variability of the frequencies with respect to time of day. Using this simple model, we attempt to establish a confidence interval of the frequencies for a new temperature profile in order to discriminate the natural variation due to temperature. Copyright (C) 1999 John Wiley & Sons, Ltd.