The proprioceptor that signals the position and movement of the first joint of crustacean legs provides an excellent system for investigating information processing and transmission in neurons that function in a graded nonspiking manner in the context of a simple motor system. The thoracic-coxal muscle receptor organ (TCMRO) spans the thoracic-coxal joint and transmits graded signals to the central nervous system via two large nonspiking axons. The response characteristics and nonlinear models of the input-output relationship for the two nonspiking TCMRO afferents (S and T fibers) were determined using white noise (Wiener kernel) methods. The best fitting linear responses of these neurons was similar, as were their second-order kernels. The gains of the afferents slowly increased with increasing frequency and reached a maximum at ~40-60Hz for the S fiber and 60-80Hz for the T fiber. Above this corner frequency, the gains of both afferents decreased at ~20dB/decade for the remainder of the 220Hz stimulus bandwidth. The shape of the first-order kernels, and hence the corresponding (linear)gain functions, of both afferents were similar when driven with different amplitudes of noise covering a 40-fold amplitude range. Predictions of the S fiber response based on the first two Wiener kernels were accurate, with the second order model producing a mean square error of 6-8%. Second-order Wiener models for the T fiber were less accurate with a mean square error of ~22-26%, but this accuracy improved to 10-16% with the incorporation of the third-order term in the Wiener expansion. The effect of cable properties on the transmission of the sensory potentials to the central nervous system was evaluated by determining the system characteristics using membrane potentials recorded 5-7mm distal to the transduction site. The major change after transmission along the axon was a low-pass filtering of the sensory signals and consequent reduction in signal bandwidth.