Rydberg-atom quantum receivers (RAQRs) enable electric-field sensing with quantum-noise-limited performance, yet their optical readout provides only magnitude measurements whose fluctuations follow Rician statistics governed by atomic projection noise, optical shot noise, reference-field injection, and short coherence times. These non-Gaussian, phase-blind measurements invalidate classical single-shot RF detectors and necessitate multi-shot quantum sensing strategies. This work develops a physically consistent multi-shot statistical model for RAQRs and derives both the optimal genie-aided likelihood-ratio test (LRT) and a practical phase-averaged LRT that removes dependence on the unknown RF-field phase. Closed-form test statistics and thresholds are obtained for both detectors, and the limits imposed by finite quantum shots-due to atomic dephasing and measurement backaction-are explicitly quantified. A fully non-coherent energy detector is also analysed, with exact detection probability derived using noncentral chi-square models. Monte Carlo results show that only 5-10 quantum shots yield major gains: the phase-averaged LRT closely approaches the genie bound and RAQR detection markedly outperforms classical RF energy detection under comparable received power. The proposed framework provides the first unified statistical basis for multi-shot Rydberg-based weak-field detection and underscores the potential of RAQRs for quantum-enhanced signal detection.