Psychophysical experiments and model simulations were performed to identify plausible neural codes representing stimulus magnitude in the Non-Pacinian I (NP I) tactile channel associated with rapidly adapting fibers. Sinusoidal mechanical displacements were applied on the fingertips of eight human subjects. The NP I channel was isolated by elevating the thresholds of the Pacinian (P) channel during forward masking. Psychophysical magnitude estimates were obtained at 40 Hz for the NP I channel and at 250 Hz for the P channel by using a small contactor (radius: 2 mm). The P channel was additionally tested with a larger contactor (radius: 4.3 mm) to compensate for the lower innervation density of the Pacinian fibers. The magnitude estimates were fitted by power functions. The exponent (1.02) obtained with the large contactor for the P channel was higher than the exponent (0.68) obtained with the small contactor, but it was not statistically different from the exponent (1.21) obtained with the small contactor for the NP I channel. This suggests that the exponent increases when more fibers are recruited in the P channel. Six hypothetical neural codes were tested by using a computational population model for the rapidly adapting afferents. The validity of each code was evaluated by comparing psychophysical and simulation exponents, by finding the correlations between the magnitude estimates and the neural code results, and by a novel distance metric for measuring the proximity between the data sets. The codes based on the number of active fibers, the total spike count, the mean and the standard deviation of the spike count distribution yielded the best results, while the codes based on the interspike intervals were not related with the magnitude estimates.