This study involves psychophysical experiments and computer simulations to investigate intensity discrimination in the non-Pacinian I (NP I) tactile channel. The simulations were based on an established population model for rapidly adapting mechanoreceptive fibers (Güuçlü & Bolanowski, 2004a). Several intensity codes were tested as decision criteria: number of active neurons, total spike count, maximal spike count, distribution of spike counts among the afferent population, and synchronization of spike times. Simulations that used the number of active fibers as the intensity code gave the most accurate results. However, the Weber fractions obtained from simulations are smaller than psychophysical Weber fractions, which suggests that only a subset of the afferent population is recruited for intensity discrimination during psychophysical experiments. Simulations could also capture the deviation from Weber's law, that is, the decrease of the Weber fraction as a function of the stimulus level, which was present in the psychophysical data. Since the psychophysical task selectively activated the NP I channel, the deviation effect is probably not due to the contribution of another tactile channel but rather is explicitly produced by the NP I channel. Moreover, because simulations with all tested intensity codes resulted in the same effect, the activity of the afferent population is sufficient to explain the deviation, without the need for a higher-order network. Depending on the intensity code used, the mechanical spread of the stimulus, rate-intensity functions of the tactile fibers, and the decreasing spike-phase jitter contribute to the deviation from Weber's law.