The timing of events within the nervous system is a critical feature of signal processing and integration. In neurotransmission, the synaptic latency, the time between stimulus delivery and appearance of the synaptic event, is generally thought to be directly related to the complexity of that pathway. In horizontal brain stem slices, we examined synaptic latency and its shock-to-shock variability (synaptic jitter) in medial nucleus tractus solitarius (NTS) neurons in response to solitary tract (ST) electrical activation. Using a visualized patch recording approach, we activated ST 1-3 mm from the recorded neuron with short trains (50-200 Hz) and measured synaptic currents under voltage clamp. Latencies ranged from 1.5 to 8.6 ms, and jitter values (SD of intraneuronal latency) ranged from 26 to 764 micros (n = 49). Surprisingly, frequency of synaptic failure was not correlated with either latency or jitter (P > 0.147; n = 49). Despite conventional expectations, no clear divisions in latency were found from the earliest arriving excitatory postsynaptic currents (EPSCs) to late pharmacologically polysynaptic responses. Shortest latency EPSCs (<3 ms) were mediated by non-N-methyl-D-aspartate (non-NMDA) glutamate receptors. Longer latency responses were a mix of excitatory and inhibitory currents including non-NMDA EPSCs and GABAa receptor-mediated currents (IPSC). All synaptic responses exhibited prominent frequency-dependent depression. In a subset of neurons, we labeled sensory boutons by the anterograde fluorescent tracer, DiA, from aortic nerve baroreceptors and then recorded from anatomically identified second-order neurons. In identified second-order NTS neurons, ST activation evoked EPSCs with short to moderate latency (1.9-4.8 ms) but uniformly minimal jitter (31 to 61 micros) that were mediated by non-NMDA receptors but had failure rates as high as 39%. These monosynaptic EPSCs in identified second-order neurons were significantly different in latency and jitter than GABAergic IPSCs (latency, 2.95 +/- 0.71 vs. 5.56 +/- 0.74 ms, mean +/- SE, P = 0.027; jitter, 42.3 +/- 6.5 vs. 416.3 +/- 94.4 micros, P = 0.013, n = 4, 6, respectively), but failure rates were similar (27.8 +/- 9.0 vs. 9.7 +/- 4.4%, P = 0.08, respectively). Such results suggest that jitter and not absolute latency or failure rate is the most reliable discriminator of mono- versus polysynaptic pathways. The results suggest that brain stem sensory pathways may differ in their principles of integration compared with cortical models and that this importantly impacts synaptic performance. The unique performance properties of the sensory-NTS pathway may reflect stronger axosomatic synaptic processing in brain stem compared with dendritically weighted models typical in cortical structures and thus may reflect very different strategies of spatio-temporal integration in this NTS region and for autonomic regulation.