We have observed clear Rabi oscillations of a weak probe in a strongly driven three-level ⌳ system in laser-cooled rubidium. When the coupling field is nonadiabatically switched on using a Pockels cell, transient probe gain without population inversion is obtained in the presence of uncoupled absorptions. Our results are supported by three-state computations. DOI: 10.1103/PhysRevA.64.055801 PACS number͑s͒: 42.50.Gy, 42.50.Md, 42.50.Hz, 32.80.Pj In the past decade there has been growing interest in the study of atomic coherence effects in multilevel atoms, especially coherent population trapping ͓1͔, electromagnetically induced transparency ͑EIT͒ ͓2͔, slow light propagation ͓3͔, and optical gain without inversion. Reviews of lasing without inversion have been given by Kocharovska and Scully ͓4͔ and more recently by Mompart and Corbalán ͓5͔. Most of this work has been in the steady-state regime and the fast picosecond pulsed regime ͓6͔, but there have been several studies of transient effects where the coherence is switched on or off rapidly in a time that is short compared to decay rates and Rabi frequencies, i.e., in the nonadiabatic regime. An early experiment by Fry et al. ͓7͔ demonstrated several transient effects in a three-level ⌳ system realized in a sodium vapor cell. Their observations included transient gain of a strong field when a weak field was switched on by a Pockels cell in the presence of incoherent pumping to the upper level, but without population inversion. Subsequently, several theoretical studies of nondegenerate three-level lambda, V, and cascade systems have predicted transient ringing of a probe beam with gain when a strong coupling beam is switched on ͓8,9͔. This predicted transient gain can occur without population inversion on the probe transition, and occurs with and without incoherent pumping. This ringing has also been explained by Vaccaro et al. ͓10͔ via stochastic wave-function diagrams. Transient ringing of a threelevel system with gain has been observed in the radiofrequency regime in nitrogen-vacancy centers in diamond samples ͓11͔ but not yet in the optical region. Since the ringing occurs at approximately the Rabi frequency of the strong field, experiments in the optical region would need to be carried out in a Doppler-free configuration or ideally in a laser-cooled sample to avoid the Doppler effects masking the coherent effects.In our previous work ͓12͔ we observed the transient approach to EIT in a cold rubidium ⌳ system in a magnetooptical trap ͑MOT͒. When the coupling field was switched on by a Pockels cell, the rapid rise in probe transparency exhibited an overshoot before settling down to the steady state. This overshoot was interpreted as the first Rabi halfcycle in the transient ringing, but it did not reach gain owing to absorption on uncoupled Zeeman transitions and twophoton dephasing effects. In the experiments reported in this paper the coupling field intensity has been increased to the point where a clear Rabi ringing cycle reaching well into gain is observ...