1. The aim of this investigation was to identify static gamma-axons which do not drive any Ia afferents at any stimulus frequency in any spindle which they supply, and to determine their occurrence in various hindlimb muscles (peroneus tertius, brevis, longus and tenuissimus). 2. Ia responses to static gamma stimulation were classified as 'non-driven' when the discharge did not follow the stimulation frequency, or its subharmonics, at any time during a linear increase in stimulus frequency up to 150 Hz lasting 2-3 s, and when tested at two muscle lengths--except in the tenuissimus muscle. In almost all experiments, cross-correlograms were used in addition to evaluate the percentage of these 'non-driven' responses in which a time-locking of discharge to stimulus pulses was obscured by irregularity of the Ia discharge. 3. In 104 spindles, out of 347 responses to stimulation of single static gamma-axons 332 (93%) could be characterized, and of these, 57% (183) were of the non-driven type. The mean number of static gamma effects characterized per spindle was 4.1 (fourteen experiments). In the large majority of spindles (79%, 82 out of 104) at least one response was of the non-driven type. 4. Of the static gamma-axons studied 16% were called 'non driving' ('ndr' gamma s-axons) because they elicited non-driven effects, and since they had the same qualitative effect consistently in all spindles whose discharge was modulated by stimulating them they were called specific 'ndr' axons. If axons with non-driven effects, but acting on one spindle were included in the 'non-driving' category the proportion was 23%. Of spindles tested 63% were innervated by at least one 'ndr' axon. 5. Absence of Ia driving during ramp frequency stimulation of gamma s-axons has been equated with selective bag2 contraction. All the non-driven responses identified in this study cannot be attributed to exclusive bag2 involvement because the total number of 'ndr' responses was too high. In fact, in the isolated spindle preparation bag2 and chain co-contraction were shown to elicit non-driven responses, so chain contraction is not detected reliably in all experimental conditions. Possibly chain fibre contraction is sometimes too weak to dominate the response, or can be of a non-driving character.(ABSTRACT TRUNCATED AT 400 WORDS)