1. Responses were recorded from both primary and secondary endings of soleus muscle spindles in the anaesthetized cat during slow stretches of the muscle after conditioning contractions at different lengths. 2. After a 5 mm length step and a fusimotor-strength contraction given at the longer length, on return to the initial length the response to a slow test stretch (0.5 mm s-1) showed a change in slope midway through the stretch which was attributed to taking up of slack in intrafusal fibres. 3. The onset of the change in slope during the test stretch depended on the size of the conditioning step. With no conditioning length change, just a fusimotor-strength contraction, there was no slope change and the response consisted of an initial burst followed by a maintained high rate of discharge. 4. Following a conditioning length step, the point of onset of the slope change during the test stretch could be altered by stimulating single identified fusimotor fibres to the spindle. Stimulating some static axons produced large changes in the stretch response while other static axons and dynamic axons had only small effects. 5. Many secondary endings showed a delay in onset of their response to a test stretch, dependent on the size of the preceding conditioning step, signalling the presence of slack in much the same way as the primary endings. Other secondary endings, however, appeared to have stretch responses that were largely independent of muscle conditioning. 6. Muscle history-independent responses of secondary endings were associated with low axonal conduction velocities. It is proposed that secondary endings which remain unaffected by muscle conditioning lie on more distal regions of nuclear chain fibres in the S2-S5 position. Here they are stimulated during both the take-up of slack and the subsequent direct stretch of the intrafusal fibres.