The contribution of Ca2+-induced Ca2+ release (CICR) to trigger muscle contraction is controversial. It was studied on isolated muscle fibres using synthetic localized increases in Ca2+ concentration, SLICs, generated by two-photon photorelease from nitrodibenzofuran (NDBF)-EGTA just outside the permeabilized plasma membrane. SLICs provided a way to increase cytosolic [Ca2+] rapidly and reversibly, up to 8 mu M, levels similar to those reached during physiological activity. They improve over previous paradigms in rate of rise, locality and reproducibility. Use of NDBF-EGTA allowed for the separate modification of resting [Ca2+], trigger [Ca2+] and resting [Mg2+]. In frog muscle, SLICs elicited propagated responses that had the characteristics of CICR. The threshold [Ca2+] for triggering a response was 0.5 mu M or less. As this value is much lower than concentrations prevailing near channels during normal activity, the result supports participation of CICR in the physiological control of contraction in amphibian muscle. As SLICs were applied outside cells, the primary stimulus was Ca2+, rather than the radiation or subproducts of photorelease. Therefore the responses qualify as 'classic' CICR. By contrast, mouse muscle fibres did not respond unless channel-opening drugs were present at substantial concentrations, an observation contrary to the physiological involvement of CICR in mammalian excitation-contraction coupling. In mouse muscle, the propagating wave had a substantially lower release flux, which together with a much higher threshold justified the absence of response when drugs were not present. The differences in flux and threshold may be ascribed to the absence of ryanodine receptor 3 (RyR3) isoforms in adult mammalian muscle.