The developing neuromuscular junction (NMJ) of Drosophila larvae is used as a simple model for genetic analysis and activity-dependent processes at glutamatergic synapses. However, due to the mostly chronic effects caused by mutations, little is known in this system about acute experience-dependent regulation of synaptic performance. We therefore used a newly developed behavioural assay (1) to stimulate NMJs by enhanced larval locomotor activity to characterise the first two hours of experience-dependent potentiation of signal transmission. Intracellular recordings of Drosophila larvae muscle 6 were performed in bridge mode and two-electrode voltage clamp (AxoClamp 2B, Axon Instruments) in haemolymph-like solution 3 solution. Data are expressed for miniature excitatory potentials (mEJP) or evoked excitatory potentials/currents (eEJP/eEJC) as mean±S.E.M., where n indicates the number of different larvae. Statistical analyses were carried out using ordinary ANOVA where applicable and a two-tailed Student’s t test for unpaired or paired data. *P<0.05 considered statistically significant. We show that after resting signal transmission (lag-phase [0-35min], mEJP: 0.98±0.07mV, n=19, eEJP: 40.5±1.8mV, n=19) experience-dependent activation of the NMJ led initially to enhanced mEJP and eEJP amplitudes during a transient phase which was driven by the release of large presynaptic quanta of glutamate (phase I [35-85min], mEJP: 1.70±0.08mV*, n=33, eEJP: 50.3±1.5mV*, n=20). This period was followed by a second phase (phase II [90-120min], mEJP: 1.05±0.08mV, n=16, eEJP: 51.6±0.9mV*, n=13) which was sensitive to specific NMDA receptor (NMDAR) antagonists. Furthermore, when analysing postsynaptic decay kinetics of eEJCs we found no change during phase II (lag-phase: τ=3.8±0.1ms, n=4 vs phase II: τ=3.0±0.3ms, n=4) nor after application of APV (NMDAR antagonist) in phase II (τ=2.9±0.5ms, n=3), indicating that NMDARs with typical decay kinetics are not postsynaptically active. To prove the presence of presynaptic Drosophila NMDARs (DNMDARs) we expressed dominant negative or wild type DNMDAR constructs post- and presynaptically, with only the expression of dominant negative DNMDARs in motorneurons eliminating the APV-sensitive component of potentiation in phase II reinforcing that previously unidentified presynaptic DNMDARs are responsible for the experience-dependent potentiation in phase II. By defining the two mechanistically distinct phases this study offers new insights into the mechanisms mediating the first two hours of experience-dependent potentiation of glutamatergic signal transmission.