Chinchillas were trained with shock-avoidance procedures to discriminate or detect pure-tone frequency differences. Inital attempts at determining frequency-difference thresholds utilized a complex discrimination paradigm and a go-no-go response. Although discriminations of large frequency differences were obtained from chinchillas after considerable training, discriminations of small frequency differences could not be trained. Therefore, determinations of frequency-difference thresholds could not be made with the complex discrimination paradigm and the go-no-go response task. A simple detection paradigm, involving the detection of frequency alternation in an ongoing train of tone bursts proved to be a more successful technique. Frequency-alternation detection was quickly learned by six chinchillas, and frequency-difference thresholds were obtained with an adaptive sequential procedure. Psychometric functions were reconstructed from the threshold tracking data of chinchillas, and comparisons were made with differential frequency thresholds from cats and humans obtained by previous investigators. Differential frequency thresholds from chinchillas paralleled those from cats and were about twice as large. Differential frequency thresholds from humans were considerably smaller than from chinchillas, especially for low-frequency tones. When a constant detectability index was used to specify differential frequency sensitiviy in chinchillas, chinchilla and human differential sensitivity functions paralleled one another. Human differential sensitivity was about ten times better than that of the chinchilla.