Accurate ice-affected streamflow records are difficult to obtain for several reasons. Problems measuring stage, variable backwater conditions, access limitations in wintertime, and problems measuring flowing water under ice cover all contribute to make ice-affected streamflow records less accurate than open-channel streamflow records. The inaccuracy of ice-affected streamflow records is particularly troublesome for small streams where Instream-Flow water rights exist. The Colorado Water Conservation Board uses these water rights to protect in-stream aquatic communities. In January and February 2002, the U.S. Geological Survey, in cooperation with the Colorado Water Conservation Board, conducted an experiment using a sodium chloride tracer to determine streamflow under ice cover. The purpose of this study is to determine the usefulness and accuracy of ice-affected streamflow records using a sodium chloride tracer that was automatically injected into the stream. The tracer was injected at two gaging stations once per day for up to 25 days. Multiple-parameter water-quality sensors at the two gaging stations monitored background and peak tracer concentrations and conductance. These data were used to determine discharge at each site. A comparison of current-meter measurements to tracer-dilution discharge measurements shows an underestimation of discharge due to inaccuracy of current-meter measurements with ice cover and inconsistent tracer-pump rates caused by partial freezing of the tracer solution in the injection lines.