This article addresses two issues. Firstly, the hypothesis that response latency might be a neural code for tone frequency was examined in single-neuron data from the primary auditory cortex of anesthetized cats. Minimal response latencies for characteristic frequency (CF) tones were independent of neural CF. Mean response latencies for a constant amplitude CF tone were also independent of CF. These data, and the fact that cortical neurons do not have an obvious independent referent for stimulus onset time, do not support the view that latency is a code for frequency. Secondly, to investigate a simple threshold model of spike initiation time, we describe the prolongations of response latency with increases in stimulus rise time and their dependence on the peak amplitude of the stimulus. These data show that in cortical neurons, it is not the peak stimulus intensity which determines first-spike latency, and second, that the response latencies are systematically not those expected on the basis of simple threshold model.