A large variability of paired-pulse plasticity (PPP) has been reported in CA3–CA1 synapses, and paired-pulse facilitation (PPF) has long been extensively used as a relative index of release probability (Pr) from the presynaptic terminal. One of the most common ways of studying Pr and PPP is to pass paired-pulse stimulation (PPS) through an electrode to fire an action potential (AP), which in turn opens Ca2+ channels in the presynaptic boutons to evoke transmitter release. However, when the postsynaptic responses were elicited by electrical stimulations, the presynaptic APs were usually not monitored. The reliability of presynaptic activation, the difference of AP firing thresholds between the first and the second pulses of PPS, and its relationship with Pr and PPP have been largely ignored. Here we show that the AP firing thresholds in the same CA3 pyramidal cells (axons) are obviously lower for the second pulses of the PPS than for the first pulses. When single (or small numbers of) presynaptic axons were stimulated by low-intensity PPS, a large variation in the postsynaptic response probability of the first pulses (Pres1) and PPP was observed. Increasing stimulation intensities resulted in the conversion of a lower Pres1 and PPF to a higher Pres1 and paired-pulse depression. These results indicate that changes in the reliability of AP initiation by the first pulse of the PPS may account for the high variability of Pres1 and PPP observed in CA1 pyramidal cells of the hippocampus, therefore PPF may not be a reliable index of Pr .