There is a controversy, to which extend cochlear stimulation with near infrared laser pulses at a wavelength of 1860 nm is based on optoacoustic stimulation of intact hair cells or -in contrast- is based on direct stimulation of the nerve cells in absence of functional hair cells. Thermal and stress confinement conditions apply, because of the pulse duration range (5 ns, 10 μs-20 ms) of the two lasers used. The dependency of the signal characteristics on pulse peak power and pulse duration was investigated in this study. The compound action potential (CAP) was measured during stimulation of the cochlea of four anaesthetized guinea pigs, which were hearing at first and afterwards acutely deafened using intracochlear neomycin-rinsing. For comparison hydrophone measurements in a water tank were performed to investigate the optoacoustic signals at different laser interaction regimes. With rising pulse peak power CAPs of the hearing animals showed first a threshold, then a positively correlated and finally a saturating dependency. CAPs also showed distinct responses at laser onset and offset separated with the pulse duration. At pulse durations shorter than physiological response times the signals merged. Basically the same signal characteristics were observed in the optoacoustic hydrophone measurements, scaled with the sensitivity and response time of the hydrophone. Taking together the qualitative correspondence in the signal response and the absence of any CAPs in deafened animals our results speak in favor of an optoacoustic stimulation of intact hair cells rather than a direct stimulation of nerve cells.