1. Quick stretches and releases were applied to small bundles of glycerinated fibres of rabbit psoas and insect fibrillar flight muscle. The resulting tension changes were recorded at various temperatures and amplitudes of length change. The results from the two preparations had many features in common. At temperatures near 0 degrees C the asymmetry of the initial tension recovery after stretch and release originally reported in living frog fibres by Huxley & Simmons (1971 alpha) was very obvious. 2. The complete tension course could be described as an elastic change occurring simultaneously with the length change followed by recovery consisting of the sum of a number of exponential terms. These terms usually corresponded to the phases discernible without curve fitting, but in some cases a monotonic rise or fall of tension was seen to consist of two components only after curve fitting. 3. After either stretch or release there was a phase of rapid tension recovery towards the value before the length change. The rate constant of this phase increased as the amplitude of stretch or release was increased to about 2 nm/half sarcomere. At higher amplitudes it remained nearly constant 4. At temperatures near 0 degrees C there was a second and much slower continuation of the recovery after stretch. The rate constant of this second phase was much more sensitive to temperature than that of the first phase and it became slower with increasing amplitude of stretch. As the temperature was raised the speed of the second phase approached the speed of the first phase so that at room temperatures the initial tension recovery after stretch and release was nearly symmetrical. 5. Under many conditions these processes were followed by a change in the opposite direction, the 'delayed tension' described by earlier workers. This third phase of tension change had about the same temperature sensitivity as the second phase of the recovery seen after stretch. The tension due to stretch activation was not maintained in rabbit muscle, resulting in a fourth possible phase, a recovery of tension towards the value before the length change. This was absent or of low amplitude in insect flight muscle. 6. We interpret these tension changes on the basis of an extension of the non-linear model described by White & Thorson (1972). The elastic tension change and the initial fast recovery are both supposed to be properties of the attached cross-bridges, whilst the slower recovery is considered to be due to the detachment of cross-bridges which happened to be attached at the instant the length change was applied. The delayed tension reflects the approach to equilibrium of the number of attached bridges, changed by an effect of muscle length on the attachment rate. The fact that the delayed tension is not maintained in rabbit psoas muscle may be due to the effect of length on attachment rate being transitory.