The present experiments investigated the signal transfer in the isolated frog muscle spindle by using pseudorandom noise (PRN) as the analytical probe. In order to guarantee that the random stimulus covered the entire dynamic range of the receptor, PRN stimuli of different intensities were applied around a constant mean length, or PRN stimuli of the same intensity were used while varying the mean length of the spindle. Subthreshold receptor potentials, local responses, and propagated action potentials were recorded simultaneously from the first Ranvier node of the afferent stem fiber, thus providing detailed insight into the spike-initiating process within a sensory receptor. Relevant features of the PRN stimulus were evaluated by a preresponse averaging technique. Up to tau = 2 ms before each action potential the encoder selected a small set of steeply rising stretch transients. A second component of the preresponse stimulus ensemble (tau = 2-5 ms) opposed the overall stretch bias. Since each steeply rising stretch transient evoked a steeply rising receptor potential that guaranteed the critical slope condition of the encoding site, this stimulus profile was most effective in initiating action potentials. The dynamic range of the muscle spindle receptor extended from resting length, L0, to about L0 + 100 microns. At the lower limit (L0) the encoding membrane was depolarized to its firing level and discharged action potentials spontaneously. When random stretches larger than the upper region of the dynamic range were applied, the spindle discharged at the maximum impulse rate and displayed no depolarization block or "overstretch" phenomenon. Random stretches applied within the dynamic range evoked regular discharge patterns that were firmly coupled to the PRN. The afferent discharge rate increased, and the precision of phase-locking improved when the intensity of the PRN stimulus was increased around a constant mean stretch; or the mean prestretch level was raised to higher values while the intensity of the PRN stimulus was kept constant. In the case when the PRN stimulus covered the entire dynamic range, the temporal pattern of the afferent discharge remained constant for at least 10 consecutive sequences of PRN. A spectral analysis of the discharge patterns averaged over several sequences of PRN was employed. At the same stimulus intensity the response spectra displayed low-pass filter characteristics with a 10-dB bandwidth of 300 Hz and a high-frequency slope of -12 dB/oct. Increasing the mean intensity of the PRN stimulus or raising the prestretch level increased the response power.(ABSTRACT TRUNCATED AT 400 WORDS)